ch4_painprocessing
2025-07-17
dirname(dirname(getwd()))[1] "/Users/h/Documents/projects_local/cue_expectancy"/Users/h/Documents/projects_local/cue_expectancyPrepare dataframe
analysis_folder <- "fmri_nps_canlab"
beh <- readr::read_tsv(file.path(main_dir, "analysis/fmri/nilearn/singletrial_rampupplateau/beh/sub-all_task-all_events.tsv"))Rows: 17400 Columns: 21
── Column specification ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Delimiter: "\t"
chr (20): sub, ses, run, runtype, cue, stimulusintensity, singletrial_fname, expectrating, expectlabel, outcomerating, outcomelabel, exp...
dbl (1): trial_index
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.dv <- "nps"
base_dir <- file.path(main_dir, "analysis/fmri/nilearn/deriv01_signature/rampup_plateau_canlab")
all_nps <- read.csv(file.path(base_dir, "signature-NPS_sub-all_runtype-pvc_event-stimulus.csv"))
all_siips <- read.csv(file.path(base_dir, "signature-SIIPS_sub-all_runtype-pain_event-stimulus.csv"))
############# NOTE: run once; concatenate all csv files ########################
# file_paths <- list.files(base_dir, pattern = "sub-.*_signature-NPSroi_.*\\.csv$", full.names = TRUE, recursive = TRUE)
# all_nps <- file_paths %>%
# map_dfr(~read_csv(.x, show_col_types = FALSE), .id = "file_path")
#
# file_paths <- list.files(base_dir, pattern = "sub-.*_roi-SIIPS.*\\.csv$", full.names = TRUE, recursive = TRUE)
# all_siips <- file_paths %>%
# map_dfr(~read_csv(.x, show_col_types = FALSE), .id = "file_path")
################################################################################
df_merge <- inner_join(beh, all_nps, by = "singletrial_fname")
df_merge <- df_merge %>%
mutate(singletrial_fname = as.character(singletrial_fname)) %>%
extract(singletrial_fname, into = c("sub", "ses", "run", "runtype", "event", "trial", "cuetype", "stimintensity"),
regex = "^(sub-\\d+)_(ses-\\d+)_(run-\\d+)_runtype-(pain|vicarious|cognitive)_event-(cue|stimulus)_trial-(\\d+)_cuetype-(high|low)(?:_stimintensity-(high|med|low))?\\.nii.gz$",
remove = FALSE)
# Adjust the dataframe to handle NA for missing stimintensity, if tidyr version doesn't support `fill`
df_merge$stimintensity[df_merge$stimintensity == ""] <- NA
data <-df_merge[df_merge$runtype == "pain" ,]
# contrast code ________________________________________________________________
data$stim <- NA; data$STIM_linear <- NA; data$STIM_quadratic <- NA;
data$CUE_high_gt_low <- NA;
data$SES_linear <- NA;data$SES_quadratic <- NA
data$stim[data$stimulusintensity == "low_stim"] <- -0.5 # social influence task
data$stim[data$stimulusintensity == "med_stim"] <- 0 # no influence task
data$stim[data$stimulusintensity == "high_stim"] <- 0.5 # no influence task
data$STIM <- factor(data$stimulusintensity)
# contrast code 1 linear
data$STIM_linear[data$stimulusintensity == "low_stim"] <- -0.5
data$STIM_linear[data$stimulusintensity == "med_stim"] <- 0
data$STIM_linear[data$stimulusintensity == "high_stim"] <- 0.5
# contrast code 2 quadratic
data$STIM_quadratic[data$stimulusintensity == "low_stim"] <- -0.33
data$STIM_quadratic[data$stimulusintensity == "med_stim"] <- 0.66
data$STIM_quadratic[data$stimulusintensity == "high_stim"] <- -0.33
# social cude contrast
data$CUE_high_gt_low[data$cue == "low_cue"] <- -0.5 # social influence task
data$CUE_high_gt_low[data$cue == "high_cue"] <- 0.5 # no influence task
data$EXPECT <- data$expectrating
data$OUTCOME <- data$outcomerating
data$SES_linear[data$ses == "ses-01"] <- -0.5
data$SES_linear[data$ses == "ses-03"] <- 0
data$SES_linear[data$ses == "ses-04"] <- 0.5
# contrast code 2 quadratic
data$SES_quadratic[data$ses == "ses-01"] <- -0.33
data$SES_quadratic[data$ses == "ses-03"] <- 0.66
data$SES_quadratic[data$ses == "ses-04"] <- -0.33
stim_con1 <- "STIM_linear"
stim_con2 <- "STIM_quadratic"
iv1 <- "CUE_high_gt_low"
dv <- "nps"
dv_keyword <- "NPS"# load dataframes ______________________________________________________________
# SIIPS <- siips %>%
# select(-X)
# NPS <- nps %>%
# select(-X)
beh <- readr::read_tsv(file.path(main_dir, "analysis/fmri/nilearn/singletrial_rampupplateau/beh/sub-all_task-all_events.tsv"))Rows: 17400 Columns: 21
── Column specification ────────────────────────────────────────────────────────────────────────────────────────────────────────────────────
Delimiter: "\t"
chr (20): sub, ses, run, runtype, cue, stimulusintensity, singletrial_fname, expectrating, expectlabel, outcomerating, outcomelabel, exp...
dbl (1): trial_index
ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.# base_dir <- "/Volumes/spacetop_projects_cue/analysis/fmri/nilearn/deriv01_signature/rampup_plateau_canlab"
NPS <- read.csv(file.path(main_dir, "analysis/fmri/nilearn/deriv01_signature/rampup_plateau_canlab/signature-NPS_sub-all_runtype-pvc_event-stimulus.csv"))
SIIPS <- read.csv(file.path(main_dir, "analysis/fmri/nilearn/deriv01_signature/rampup_plateau_canlab/signature-SIIPS_sub-all_runtype-pain_event-stimulus.csv"))
# merge the three dataframes ___________________________________________________
df_merge1 <- inner_join(beh, NPS, by = "singletrial_fname")
dfmerge <- inner_join(df_merge1, SIIPS, by = "singletrial_fname")
# df_merge <- inner_join(beh, all_nps, by = "singletrial_fname")
dfmerge <- dfmerge %>%
mutate(singletrial_fname = as.character(singletrial_fname)) %>%
extract(singletrial_fname, into = c("sub", "ses", "run", "runtype", "event", "trial", "cuetype", "stimintensity"),
regex = "^(sub-\\d+)_(ses-\\d+)_(run-\\d+)_runtype-(pain|vicarious|cognitive)_event-(cue|stimulus)_trial-(\\d+)_cuetype-(high|low)(?:_stimintensity-(high|med|low))?\\.nii.gz$",
remove = FALSE)
dfmerge$NPS <- dfmerge$nps
# create folder to save data ___________________________________________________
analysis_dir <- "/Users/h/Documents/projects_local/cue_expectancy/resources/plots_dissertation/ch4"
dir.create(analysis_dir,
showWarnings = FALSE,
recursive = TRUE)
savedir <- analysis_dir
lowcuehex <- "#4274AD" # "#575a5e"
highcuehex <- "#C5263A" #"#FEAE00"data <-dfmerge[dfmerge$runtype == "pain" ,]
# contrast code ________________________________________________________________
data$stim <- NA; data$STIM_linear <- NA; data$STIM_quadratic <- NA;
data$CUE_high_gt_low <- NA;
data$SES_linear <- NA;data$SES_quadratic <- NA
data$stim[data$stimulusintensity == "low_stim"] <- -0.5 # social influence task
data$stim[data$stimulusintensity == "med_stim"] <- 0 # no influence task
data$stim[data$stimulusintensity == "high_stim"] <- 0.5 # no influence task
data$STIM <- factor(data$stimulusintensity)
data$SUB <- factor(data$sub)
# undo string format
data$outcomerating_impute[data$outcomerating_impute == "n/a"] <- NA
data$outcomerating_impute <- as.numeric(data$outcomerating_impute)
data$expectrating_impute[data$expectrating_impute == "n/a"] <- NA
data$expectrating_impute <- as.numeric(data$expectrating_impute)
# contrast code 1 linear
data$STIM_linear[data$stimulusintensity == "low_stim"] <- -0.5
data$STIM_linear[data$stimulusintensity == "med_stim"] <- 0
data$STIM_linear[data$stimulusintensity == "high_stim"] <- 0.5
# contrast code 2 quadratic
data$STIM_quadratic[data$stimulusintensity == "low_stim"] <- -0.33
data$STIM_quadratic[data$stimulusintensity == "med_stim"] <- 0.66
data$STIM_quadratic[data$stimulusintensity == "high_stim"] <- -0.33
# social cude contrast
data$CUE_high_gt_low[data$cue == "low_cue"] <- -0.5 # social influence task
data$CUE_high_gt_low[data$cue == "high_cue"] <- 0.5 # no influence task
data$EXPECT <- data$expectrating
data$OUTCOME <- data$outcomerating
data$SES_linear[data$ses == "ses-01"] <- -0.5
data$SES_linear[data$ses == "ses-03"] <- 0
data$SES_linear[data$ses == "ses-04"] <- 0.5
# contrast code 2 quadratic
data$SES_quadratic[data$ses == "ses-01"] <- -0.33
data$SES_quadratic[data$ses == "ses-03"] <- 0.66
data$SES_quadratic[data$ses == "ses-04"] <- -0.33
stim_con1 <- "STIM_linear"
stim_con2 <- "STIM_quadratic"
iv1 <- "CUE_high_gt_low"
subjects_with_inadequate_data <- data %>%
group_by(sub, CUE_high_gt_low, STIM_linear) %>% #SES_linear,
dplyr::summarise(count = n(), .groups = 'drop') %>%
filter(count < 3) %>%
distinct(sub) %>%
pull(sub)
df_filter <- data %>%
filter(!(sub %in% subjects_with_inadequate_data))
print(sprintf("after filtering out subjects that have less than 3 trials in cell, we have N=%d -> N=%d",length(unique(data$sub)), length(unique(df_filter$sub)) ))[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"1 Beh
Beh linear analysis
model.beh <- lmer(outcomerating_impute ~ CUE_high_gt_low*STIM_linear +
CUE_high_gt_low*STIM_quadratic + (CUE_high_gt_low+STIM_linear|sub), data = df_filter)
summary(model.beh)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: outcomerating_impute ~ CUE_high_gt_low * STIM_linear + CUE_high_gt_low *
STIM_quadratic + (CUE_high_gt_low + STIM_linear | sub)
Data: df_filter
REML criterion at convergence: 44907.5
Scaled residuals:
Min 1Q Median 3Q Max
-4.8573 -0.5565 -0.0002 0.5477 4.2955
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 814.63 28.542
CUE_high_gt_low 59.51 7.715 -0.03
STIM_linear 146.40 12.100 0.40 0.07
Residual 401.05 20.026
Number of obs: 5014, groups: sub, 92
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 64.0388 2.9917 90.8850 21.406 < 2e-16 ***
CUE_high_gt_low 8.1019 1.0004 86.6825 8.099 3.19e-12 ***
STIM_linear 29.9137 1.4594 91.8897 20.498 < 2e-16 ***
STIM_quadratic 0.8294 0.6058 4739.6474 1.369 0.17101
CUE_high_gt_low:STIM_linear -0.5183 1.3873 4741.0874 -0.374 0.70874
CUE_high_gt_low:STIM_quadratic -3.7623 1.2116 4740.0763 -3.105 0.00191 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) CUE_h__ STIM_l STIM_q CUE_hgh_gt_lw:STIM_l
CUE_hgh_gt_ -0.028
STIM_linear 0.348 0.055
STIM_qudrtc 0.000 -0.001 -0.001
CUE_hgh_gt_lw:STIM_l 0.001 0.001 0.001 -0.004
CUE_hgh_gt_lw:STIM_q 0.000 -0.001 -0.002 0.002 -0.001 sjPlot::tab_model(model.beh,
title = "Multilevel-modeling: \nlmer(Outcome ~ CUE * STIM + (CUE + STIM | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| outcomerating_impute | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 64.04 | 58.17 – 69.90 | <0.001 |
| CUE high gt low | 8.10 | 6.14 – 10.06 | <0.001 |
| STIM linear | 29.91 | 27.05 – 32.77 | <0.001 |
| STIM quadratic | 0.83 | -0.36 – 2.02 | 0.171 |
|
CUE high gt low × STIM linear |
-0.52 | -3.24 – 2.20 | 0.709 |
|
CUE high gt low × STIM quadratic |
-3.76 | -6.14 – -1.39 | 0.002 |
| Random Effects | |||
| σ2 | 401.05 | ||
| τ00 sub | 814.63 | ||
| τ11 sub.CUE_high_gt_low | 59.51 | ||
| τ11 sub.STIM_linear | 146.40 | ||
| ρ01 | -0.03 | ||
| 0.40 | |||
| ICC | 0.68 | ||
| N sub | 92 | ||
| Observations | 5014 | ||
| Marginal R2 / Conditional R2 | 0.117 / 0.718 | ||
equatiomatic::extract_eq(model.beh)\[ \begin{aligned} \operatorname{outcomerating\_impute}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2j[i]}(\operatorname{STIM\_linear}) + \beta_{3}(\operatorname{STIM\_quadratic}) + \beta_{4}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{5}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{2j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{2j}} \end{aligned} \end{array} \right) , \left( \begin{array}{ccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{2j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{2j}} \\ \rho_{\beta_{2j}\alpha_{j}} & \rho_{\beta_{2j}\beta_{1j}} & \sigma^2_{\beta_{2j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
Beh lineplot
dv <- "outcomerating_impute"
taskname <- "pain"
subject <- "sub"
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name <- NA; df_filter$stim_name <- NA
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <-
"low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
# DATA$levels_ordered <- factor(DATA$param_stimulus_type, levels=c("low", "med", "high"))
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
stimcue_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2), dv)
stimcue_groupwise <- summarySEwithin(
data = stimcue_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
stimcue_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "Outcome rating"
ylim <- c(-10, 60)
dv_keyword <- "outcomerating"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)
subset <- stimcue_groupwise[stimcue_groupwise$cue_ordered == "low cue",]
g <- plot_lineplot_twofactor(subset,
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex), ggtitle = "Within pain task: Outcome rating",
xlab = "Stimulus intensity", ylab = "Outcome rating")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim(42, 87)
ggsave(file.path(analysis_dir, paste0("beh_maineffect", as.character(Sys.Date()), "_lowcue.svg")), plot=g, w=4, h=4, dpi=300)
k <- cueR::plot_lineplot_twofactor_subset(stimcue_groupwise,
taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex,"high cue" = highcuehex),
ggtitle = paste0("Beh N =", length(unique(stimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "Outcome rating")Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`k<- k+ theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim(42, 87)
k
ggsave(file.path(analysis_dir, paste0("beh_maineffect", as.character(Sys.Date()), ".svg")), plot=k, w=4, h=4, dpi=300)2 NPS (df_filter)
NPS linear analysis
model.NPS <- lmer(NPS ~ CUE_high_gt_low*STIM_linear +
CUE_high_gt_low*STIM_quadratic + (CUE_high_gt_low+STIM_linear|sub), data = df_filter)boundary (singular) fit: see help('isSingular')summary(model.NPS)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: NPS ~ CUE_high_gt_low * STIM_linear + CUE_high_gt_low * STIM_quadratic + (CUE_high_gt_low + STIM_linear | sub)
Data: df_filter
REML criterion at convergence: 28051.3
Scaled residuals:
Min 1Q Median 3Q Max
-7.4553 -0.5252 -0.0012 0.5343 4.7549
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 3.81734 1.9538
CUE_high_gt_low 0.04399 0.2097 -0.05
STIM_linear 0.20085 0.4482 0.84 0.49
Residual 14.71951 3.8366
Number of obs: 5029, groups: sub, 92
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 5.1782 0.2120 91.2599 24.430 < 2e-16 ***
CUE_high_gt_low -0.1654 0.1107 203.6781 -1.495 0.1365
STIM_linear 1.2624 0.1409 191.0483 8.958 2.96e-16 ***
STIM_quadratic 0.2813 0.1159 4858.4992 2.427 0.0152 *
CUE_high_gt_low:STIM_linear -0.2640 0.2653 4860.5313 -0.995 0.3196
CUE_high_gt_low:STIM_quadratic -0.3100 0.2318 4858.7496 -1.338 0.1810
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) CUE_h__ STIM_l STIM_q CUE_hgh_gt_lw:STIM_l
CUE_hgh_gt_ -0.011
STIM_linear 0.272 0.041
STIM_qudrtc 0.000 -0.001 -0.001
CUE_hgh_gt_lw:STIM_l 0.002 0.002 0.002 -0.004
CUE_hgh_gt_lw:STIM_q 0.000 -0.002 -0.004 0.000 -0.002
optimizer (nloptwrap) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')sjPlot::tab_model(model.NPS,
title = "Multilevel-modeling: \nlmer(NPS ~ CUE * STIM + (CUE + STIM | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| NPS | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 5.18 | 4.76 – 5.59 | <0.001 |
| CUE high gt low | -0.17 | -0.38 – 0.05 | 0.135 |
| STIM linear | 1.26 | 0.99 – 1.54 | <0.001 |
| STIM quadratic | 0.28 | 0.05 – 0.51 | 0.015 |
|
CUE high gt low × STIM linear |
-0.26 | -0.78 – 0.26 | 0.320 |
|
CUE high gt low × STIM quadratic |
-0.31 | -0.76 – 0.14 | 0.181 |
| Random Effects | |||
| σ2 | 14.72 | ||
| τ00 sub | 3.82 | ||
| τ11 sub.CUE_high_gt_low | 0.04 | ||
| τ11 sub.STIM_linear | 0.20 | ||
| ρ01 | -0.05 | ||
| 0.84 | |||
| ICC | 0.21 | ||
| N sub | 92 | ||
| Observations | 5029 | ||
| Marginal R2 / Conditional R2 | 0.016 / 0.220 | ||
equatiomatic::extract_eq(model.NPS)\[ \begin{aligned} \operatorname{NPS}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2j[i]}(\operatorname{STIM\_linear}) + \beta_{3}(\operatorname{STIM\_quadratic}) + \beta_{4}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{5}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{2j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{2j}} \end{aligned} \end{array} \right) , \left( \begin{array}{ccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{2j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{2j}} \\ \rho_{\beta_{2j}\alpha_{j}} & \rho_{\beta_{2j}\beta_{1j}} & \sigma^2_{\beta_{2j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
sink("/Users/h/Desktop/NPS_output.txt") # Redirects output to 'my_output.txt'
summary(model.NPS)sink(NULL) #lineplot
dv <- "NPS"
taskname <- "pain"
subject <- "sub"
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <-
"low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
# DATA$levels_ordered <- factor(DATA$param_stimulus_type, levels=c("low", "med", "high"))
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
NPSstimcue_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2), dv)
NPSstimcue_groupwise <- summarySEwithin(
data = NPSstimcue_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
NPSstimcue_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "NPS"
ylim <- c(-10, 60)
dv_keyword <- "nps"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)
NPSsubset <- NPSstimcue_groupwise[NPSstimcue_groupwise$cue_ordered == "low cue",]
g <- plot_lineplot_twofactor(NPSsubset,
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex), ggtitle = paste0("NPS N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPS")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim( 3,6)
ggsave(file.path(analysis_dir, paste0("NPS_maineffect", as.character(Sys.Date()), "_lowcue.svg")), plot=g, w=4, h=4, dpi=300)
k <- cueR::plot_lineplot_twofactor_subset(NPSstimcue_groupwise, taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex,"high cue" = highcuehex),
ggtitle = paste0("NPS N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPS")Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`k<- k+ theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim( 4,6)
k
ggsave(file.path(analysis_dir, paste0("NPS_maineffect", as.character(Sys.Date()), ".svg")), plot=k, w=4, h=4, dpi=300)NPS pos (df_filter)
dv <- "npspos"
taskname <- "pain"
subject <- "sub"
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <-
"low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
# DATA$levels_ordered <- factor(DATA$param_stimulus_type, levels=c("low", "med", "high"))
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
NPSstimcue_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2), dv)
NPSstimcue_groupwise <- summarySEwithin(
data = NPSstimcue_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
NPSstimcue_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "NPSpos"
ylim <- c(-10, 60)
dv_keyword <- "nps"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)
NPSsubset <- NPSstimcue_groupwise[NPSstimcue_groupwise$cue_ordered == "low cue",]
g <- plot_lineplot_twofactor(NPSsubset,
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex), ggtitle = paste0("NPS N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPS")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim( 3,6)
ggsave(file.path(analysis_dir, paste0("NPSpos_maineffect", as.character(Sys.Date()), "_lowcue.svg")), plot=g, w=4, h=4, dpi=300)Warning: Removed 3 rows containing missing values or values outside the scale range (`geom_line()`).Warning: Removed 3 rows containing missing values or values outside the scale range (`geom_point()`).
Removed 3 rows containing missing values or values outside the scale range (`geom_point()`).k <- cueR::plot_lineplot_twofactor_subset(NPSstimcue_groupwise,taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex,"high cue" = highcuehex),
ggtitle = paste0("NPS (pos) N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPS")Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`k<- k+ theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2)
k
ggsave(file.path(analysis_dir, paste0("NPSpos_maineffect", as.character(Sys.Date()), ".svg")), plot=k, w=4, h=4, dpi=300)NPS neg
dv <- "npsneg"
taskname <- "pain"
subject <- "sub"
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <-
"low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
# DATA$levels_ordered <- factor(DATA$param_stimulus_type, levels=c("low", "med", "high"))
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
NPSstimcue_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2), dv)
NPSstimcue_groupwise <- summarySEwithin(
data = NPSstimcue_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
NPSstimcue_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "NPSneg"
ylim <- c(-10, 60)
dv_keyword <- "nps"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)
NPSsubset <- NPSstimcue_groupwise[NPSstimcue_groupwise$cue_ordered == "low cue",]
g <- plot_lineplot_twofactor(NPSsubset,
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex), ggtitle = paste0("NPS N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPSneg")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim( 3,6)
ggsave(file.path(analysis_dir, paste0("NPSneg_maineffect", as.character(Sys.Date()), "_lowcue.svg")), plot=g, w=4, h=4, dpi=300)Warning: Removed 3 rows containing missing values or values outside the scale range (`geom_line()`).Warning: Removed 3 rows containing missing values or values outside the scale range (`geom_point()`).
Removed 3 rows containing missing values or values outside the scale range (`geom_point()`).k <- cueR::plot_lineplot_twofactor_subset(NPSstimcue_groupwise, taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex,"high cue" = highcuehex),
ggtitle = paste0("NPSneg N =", length(unique(NPSstimcue_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "NPSneg")Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`k<- k+ theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2)
k
ggsave(file.path(analysis_dir, paste0("NPSneg_maineffect", as.character(Sys.Date()), ".svg")), plot=k, w=4, h=4, dpi=300)3 PE (PEmerge)
PEdf <- read.csv("/Users/h/Documents/projects_local/cue_expectancy/data/RL/modelfit_072024/table_pain.csv")
clean_filename <- function(filename) {
# Remove _cue that follows cuetype-
filename <- gsub("(cuetype-[^_]+)_cue", "\\1", filename)
# Remove _stim_stim that follows stimintensity-
filename <- gsub("(stimintensity-[^_]+)_stim_stim", "\\1", filename)
return(filename)
}
# Apply the function to the singletrial_fname column
PEdf$singletrial_fname <- sapply(PEdf$singletrial_fname, clean_filename)
PEmerge<- inner_join(df_filter, PEdf, by = "singletrial_fname")
#
# PEmerge <- merge(dfmerge, PEdf, by = c("src_subject_id", "session_id", "param_run_num", "trial_index"))
print(PEmerge)# A tibble: 4,937 × 133
trial_index.x cue stimulusintensity singletrial_fname sub ses.x run runtype.x event trial cuetype stimintensity expectrating.x
<dbl> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr> <chr>
1 1 low_cue med_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 000 low med n/a
2 2 high_cue low_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 001 high low 116.36
3 3 high_cue high_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 002 high high 38.16
4 4 low_cue med_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 003 low med 34.39
5 5 high_cue low_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 004 high low 80.5
6 6 low_cue high_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 005 low high 7.53
7 7 low_cue low_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 006 low low 29.77
8 8 high_cue med_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 007 high med 54.46
9 9 low_cue high_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 008 low high 26.26
10 10 high_cue med_stim sub-0005_ses-03_ru… sub-… ses-… run-… pain stim… 009 high med 65.56
# ℹ 4,927 more rows
# ℹ 120 more variables: expectlabel.x <chr>, outcomerating.x <chr>, outcomelabel.x <chr>, expectrating_impute <dbl>,
# expectlabel_impute.x <chr>, outcomerating_impute <dbl>, outcomelabel_impute.x <chr>, expectrating_RT.x <chr>,
# expectrating_RTimpute.x <chr>, outcomerating_RT.x <chr>, outcomerating_RTimpute.x <chr>, pain_stimulus_delivery_success.x <chr>,
# X.x <int>, file_path.x <int>, nps <dbl>, nps_corr <dbl>, nps_cosine <dbl>, npspos <dbl>, npspos_corr <dbl>, npspos_cosine <dbl>,
# npsneg <dbl>, npsneg_corr <dbl>, npsneg_cosine <dbl>, pos_nps_vermis <dbl>, pos_nps_vermis_corr <dbl>, pos_nps_vermis_cosine <dbl>,
# pos_nps_rIns <dbl>, pos_nps_rIns_corr <dbl>, pos_nps_rIns_cosine <dbl>, pos_nps_rV1 <dbl>, pos_nps_rV1_corr <dbl>, …PE.model <- lmer(nps ~ PE_mdl2 + (1 | sub), data=PEmerge, control = lmerControl(optimizer = "nmkbw"))
summary(PE.model)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: nps ~ PE_mdl2 + (1 | sub)
Data: PEmerge
Control: lmerControl(optimizer = "nmkbw")
REML criterion at convergence: 27628.2
Scaled residuals:
Min 1Q Median 3Q Max
-7.1811 -0.5308 -0.0090 0.5348 4.8091
Random effects:
Groups Name Variance Std.Dev.
sub (Intercept) 3.874 1.968
Residual 15.008 3.874
Number of obs: 4937, groups: sub, 88
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 5.140e+00 2.182e-01 8.715e+01 23.552 < 2e-16 ***
PE_mdl2 1.659e-02 2.346e-03 4.904e+03 7.071 1.76e-12 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr)
PE_mdl2 -0.048sjPlot::tab_model(PE.model,
title = "Multilevel-modeling: \nlmer(NPS ~ PE + (PE | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| nps | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 5.14 | 4.71 – 5.57 | <0.001 |
| PE mdl2 | 0.02 | 0.01 – 0.02 | <0.001 |
| Random Effects | |||
| σ2 | 15.01 | ||
| τ00 sub | 3.87 | ||
| ICC | 0.21 | ||
| N sub | 88 | ||
| Observations | 4937 | ||
| Marginal R2 / Conditional R2 | 0.009 / 0.212 | ||
equatiomatic::extract_eq(PE.model)\[ \begin{aligned} \operatorname{nps}_{i} &\sim N \left(\alpha_{j[i]} + \beta_{1}(\operatorname{PE\_mdl2}), \sigma^2 \right) \\ \alpha_{j} &\sim N \left(\mu_{\alpha_{j}}, \sigma^2_{\alpha_{j}} \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
sink("/Users/h/Desktop/PEnps_output.txt") # Redirects output to 'my_output.txt'
summary(PE.model)sink(NULL) #
library(ggplot2)
min_value <- -40
max_value <- 40
PEmerge$sub <- factor(PEmerge$sub)
plot.PE_NPS <- ggplot(PEmerge, aes(x = PE_mdl2, y = NPS)) +
geom_point(aes(colour = sub), size = .1) + # Points colored by subject
geom_smooth(aes(colour = sub), method = 'lm', formula = y ~ x, se = FALSE, size = .3, linetype = "dashed") + # Subject-wise regression lines
geom_smooth(method = 'lm', formula = y ~ x, se = FALSE, size = .5, color = "black") + # Group regression line
# ylim(min_value, max_value) + # Set y-axis limits
# xlim(min_value, max_value)
theme_classic2() +# Use a theme with a white background
xlab("PE") +
theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 24, ), # Axis titles
axis.text = element_text(size = 18), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5),# Plot title
legend.position = "none"
)
plot.PE_NPS
ggsave("/Users/h/Documents/projects_local/cue_expectancy/resources/plots_dissertation/ch4/ch4_NPSPE.png",plot.PE_NPS, w=5, h=4, dpi=300)Q. How correlated are the outcome ratings and NPS values?? TODO: can we indicate correlation based on significance
library(corrplot)
PEmerge$cuetype_numeric <- as.numeric(as.factor(PEmerge$cuetype))
PEmerge$beh_PE <- PEmerge$outcomerating_impute - PEmerge$expectrating_impute
subset <- PEmerge[, c( "outcomerating_impute", "expectrating_impute", "PE_mdl2","beh_PE", "nps", "npspos", "npsneg", "cuetype_numeric") ]
cor_matrix <- cor(subset, use = "pairwise.complete.obs")
corrplot(cor_matrix, method = "circle")
# Correlation between cuetype and model-based PE
correlation_model <- cor_matrix["cuetype_numeric", "PE_mdl2"]
print(correlation_model)[1] 0.611056# Correlation between cuetype and behavioral PE
correlation_behavioral <- cor_matrix["cuetype_numeric", "beh_PE"]
print(correlation_behavioral)[1] 0.4052953Tor suggestion lmer with PE
What to do when model fit is singular https://stackoverflow.com/questions/54597496/how-to-cope-with-a-singular-fit-in-a-linear-mixed-model-lme4 Singular. it means * one or more random effect variances are near zero, * and the model is overfitting or overparameterized (can’t estimate all the random slopes reliably). suggestion. Simplify the random effect structure
Q. Why would i have a singular random effets structure? A1. when you have too many variables you’re fitting. You include too many random slopes relative to the amount of data A2. the predictors are highly collinear (PE is similar to cue)
NPS pos is predicted by cue and stim Once PE comes into the picture, matrix is singular
summary(NPSPE)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: NPS ~ STIM_linear + STIM_quadratic + PE_mdl2 + (1 + PE_mdl2 | sub)
Data: PEmerge
REML criterion at convergence: 27580.9
Scaled residuals:
Min 1Q Median 3Q Max
-7.3591 -0.5302 -0.0024 0.5372 4.7212
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 3.918e+00 1.9795146
PE_mdl2 2.835e-07 0.0005324 -1.00
Residual 1.485e+01 3.8539140
Number of obs: 4937, groups: sub, 88
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 5.183e+00 2.194e-01 8.666e+01 23.620 < 2e-16 ***
STIM_linear 1.058e+00 1.544e-01 4.842e+03 6.854 8.1e-12 ***
STIM_quadratic 2.661e-01 1.175e-01 4.846e+03 2.265 0.02358 *
PE_mdl2 7.495e-03 2.680e-03 4.855e+03 2.796 0.00519 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) STIM_l STIM_q
STIM_linear 0.028
STIM_qudrtc 0.001 0.014
PE_mdl2 -0.076 -0.491 -0.031
optimizer (nloptwrap) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')#
# model.NPS <- lmer(NPS ~ CUE_high_gt_low*STIM_linear +
# CUE_high_gt_low*STIM_quadratic + (CUE_high_gt_low+STIM_linear|sub), data = df_filter)
NPSPE <- lmer(NPS ~ STIM_linear +
STIM_quadratic + PE_mdl2 + (1 + PE_mdl2 |sub),
data = PEmerge)boundary (singular) fit: see help('isSingular') # control=lmerControl(isSingular(NPSPE, tol = 1e-4)))
# check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
summary(NPSPE)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: NPS ~ STIM_linear + STIM_quadratic + PE_mdl2 + (1 + PE_mdl2 | sub)
Data: PEmerge
REML criterion at convergence: 27580.9
Scaled residuals:
Min 1Q Median 3Q Max
-7.3591 -0.5302 -0.0024 0.5372 4.7212
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 3.918e+00 1.9795146
PE_mdl2 2.835e-07 0.0005324 -1.00
Residual 1.485e+01 3.8539140
Number of obs: 4937, groups: sub, 88
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 5.183e+00 2.194e-01 8.666e+01 23.620 < 2e-16 ***
STIM_linear 1.058e+00 1.544e-01 4.842e+03 6.854 8.1e-12 ***
STIM_quadratic 2.661e-01 1.175e-01 4.846e+03 2.265 0.02358 *
PE_mdl2 7.495e-03 2.680e-03 4.855e+03 2.796 0.00519 **
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) STIM_l STIM_q
STIM_linear 0.028
STIM_qudrtc 0.001 0.014
PE_mdl2 -0.076 -0.491 -0.031
optimizer (nloptwrap) convergence code: 0 (OK)
boundary (singular) fit: see help('isSingular')sjPlot::tab_model(NPSPE,
title = "model 2: \nlmer(NPS ~ cue + stim + PE + (cue + stim | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| NPS | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 5.18 | 4.75 – 5.61 | <0.001 |
| STIM linear | 1.06 | 0.76 – 1.36 | <0.001 |
| STIM quadratic | 0.27 | 0.04 – 0.50 | 0.024 |
| PE mdl2 | 0.01 | 0.00 – 0.01 | 0.005 |
| Random Effects | |||
| σ2 | 14.85 | ||
| τ00 sub | 3.92 | ||
| τ11 sub.PE_mdl2 | 0.00 | ||
| ρ01 sub | -1.00 | ||
| N sub | 88 | ||
| Observations | 4937 | ||
| Marginal R2 / Conditional R2 | 0.021 / NA | ||
equatiomatic::extract_eq(NPSPE)\[ \begin{aligned} \operatorname{NPS}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1}(\operatorname{STIM\_linear}) + \beta_{2}(\operatorname{STIM\_quadratic}) + \beta_{3j[i]}(\operatorname{PE\_mdl2}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{3j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{3j}} \end{aligned} \end{array} \right) , \left( \begin{array}{cc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{3j}} \\ \rho_{\beta_{3j}\alpha_{j}} & \sigma^2_{\beta_{3j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
library(sjPlot)
# Plot marginal effect (partial slope) for PE
library(ggplot2)
# Plot fitted values by subject
PEmerge$predicted_NPS <- predict(NPSPE)
# ggplot(PEmerge, aes(x = PE_mdl2, y = predicted_NPS, color = sub)) +
# geom_line(aes(group = sub), alpha = 0.4) +
# labs(title = "Subject-specific predicted slopes for Prediction Error",
# x = "Prediction Error (PE_mdl2)",
# y = "Predicted NPS") +
# theme_minimal(base_size = 14) +
# theme(legend.position = "none")
library(dplyr)
PEmerge_sorted <- PEmerge %>%
arrange(sub, PE_mdl2)
ggplot(PEmerge_sorted, aes(x = PE_mdl2, y = predicted_NPS, color = sub)) +
geom_line(aes(group = sub), alpha = 0.4) +
labs(title = "Subject-specific predicted slopes for Prediction Error",
x = "Prediction Error (PE_mdl2)",
y = "Predicted NPS") +
theme_minimal(base_size = 14) +
theme(legend.position = "none")
library(ggeffects)
library(ggplot2)
# Get predicted values for PE across its range
pe_pred <- ggpredict(NPSPE, terms = "PE_mdl2")
# Plot
ggplot(pe_pred, aes(x = x, y = predicted)) +
geom_line() +
geom_ribbon(aes(ymin = conf.low, ymax = conf.high), alpha = 0.2) +
labs(title = "Partial Effect of Prediction Error on NPS",
x = "Prediction Error (PE_mdl2)",
y = "Predicted NPS") +
theme_minimal(base_size = 14)
plot_model(NPSPE, type = "eff", terms = c( "STIM_linear","PE_mdl2"))
# # model 2: NPS PE positive values
# NPSPEpos <- lmer(npspos~ cuetype + stimintensity + PE_mdl2 + (cuetype + stimintensity|sub),
# data = PEmerge,
# control=lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
# summary(NPSPEpos)
# vif(NPSPEpos)
# sjPlot::tab_model(NPSPEpos,
# title = "model 2: \nlmer(NPSpos ~ cue + stim + PE + (cue + stim | sub), data = pain)",
# CSS = list(css.table = '+font-size: 12;'))
# equatiomatic::extract_eq(NPSPEpos)4 NPS nociceptive ROI (df_filter.NA)
# STEP 0. filter data __________________________________________________________
# Make sure that each condition cell has adequate amount of trials
dependent_vars <- c("nps","npspos","npsneg",
"pos_nps_vermis","pos_nps_rIns","pos_nps_rV1","pos_nps_rThal",
"pos_nps_lIns","pos_nps_rdpIns","pos_nps_rS2_Op","pos_nps_dACC",
"neg_nps_rLOC","neg_nps_lLOC","neg_nps_rpLOC","neg_nps_pgACC","neg_nps_lSTS","neg_nps_rIPL","neg_nps_PCC" )
newlist <- c("NPS","NPS positive weights","NPS negative weights",
"Vermis","rIns","rV1","rThal",
"lIns","rdpIns","rS2 & Op","dACC",
"neg_nps_rLOC","neg_nps_lLOC","neg_nps_rpLOC","neg_nps_pgACC","neg_nps_lSTS","neg_nps_rIPL","neg_nps_PCC" )
# Initialize a DataFrame to collect combined results
combined_se_calc_cooksd <- data.frame()
for (i in seq_along(dependent_vars)) {
dv <- dependent_vars[i]
dv_title <- newlist[i]
print(paste("_______",i, dv, "_______"))
subjects_with_inadequate_data <- data %>%
group_by(sub, CUE_high_gt_low, STIM_linear) %>% #SES_linear,
dplyr::summarise(count = n(), .groups = 'drop') %>%
filter(count < 3) %>%
distinct(sub) %>%
pull(sub)
df_filter <- data %>%
filter(!(sub %in% subjects_with_inadequate_data))
print(sprintf("after filtering out subjects that have less than 3 trials in cell, we have N=%d -> N=%d",length(unique(data$sub)), length(unique(df_filter$sub)) ))
## QC. check NPS distribution __________________________________________________
df_filter.NA <- df_filter %>% filter(!is.na(nps)) # Remove NA values
head(df_filter.NA)
combined_se_calc_cooksd <- data.frame()
taskname = "pain"
ggtitle <- paste(taskname, " - NPS (degree)")
title <- paste(taskname, " - actual")
subject <- "sub"
w <- 10
h <- 6
analysis_dir <-
file.path(main_dir,
"analysis",
"mixedeffect",
analysis_folder,
as.character(Sys.Date())) # nolint
dir.create(analysis_dir,
showWarnings = FALSE,
recursive = TRUE)
savedir <- analysis_dir
df_filter.NA <- df_filter.NA %>%
group_by(sub) %>%
mutate(NPS_scaled = scale(nps)) %>%
ungroup()
# 1. rename reordering for plots ______________________________________________
df_filter.NA$cue_name <- NA; df_filter.NA$stim_name <- NA
df_filter.NA$cue_name[df_filter.NA$cue == "high_cue"] <- "high cue"
df_filter.NA$cue_name[df_filter.NA$cue == "low_cue"] <- "low cue"
df_filter.NA$stim_name[df_filter.NA$stimulusintensity == "high_stim"] <- "High"
df_filter.NA$stim_name[df_filter.NA$stimulusintensity == "med_stim"] <- "Med"
df_filter.NA$stim_name[df_filter.NA$stimulusintensity == "low_stim"] <- "Low"
df_filter.NA$stim_ordered <- factor(df_filter.NA$stim_name, levels = c("Low", "Med", "High"))
df_filter.NA$cue_ordered <- factor(df_filter.NA$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# 2. lmer model ________________________________________________________________
model_savefname <- file.path(
analysis_dir,
paste(
"lmer_nps-",dv,"_",as.character(Sys.Date()),".txt",
sep = ""
)
)
formula_string <- paste(dv, "~ CUE_high_gt_low * STIM_linear + CUE_high_gt_low * STIM_quadratic + (CUE_high_gt_low + STIM_linear | sub)")
model_formula <- as.formula(formula_string)
# Fit the model
sink(model_savefname)
# This control parameter which ignores the singular warning allows for a zero estimate.
# preoceed with caution. # Read: https://stackoverflow.com/questions/54597496/how-to-cope-with-a-singular-fit-in-a-linear-mixed-model-lme4
model.npscuestim <- lmer(model_formula, data = df_filter.NA,
control=lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
# control = lmerControl(optimizer = "nmkbw"))
print(summary(model.npscuestim))
sink()
# Check why convergence is negative ____________________________________________
aa <- lme4::allFit(model.npscuestim)
opt.summary <- glance(aa) |> select(optimizer, AIC, NLL_rel) |> arrange(NLL_rel)
print(opt.summary)
# R2 ___________________________________________________________________________
r_squared <- MuMIn::r.squaredGLMM(model.npscuestim)
R2m = r_squared[1] # Marginal R-squared
R2c = r_squared[2] # Conditional R-squared
# summary statistics calculate mean and se ____________________________________
NPSstimcue_subjectwise <- meanSummary(df_filter.NA,
c(subject, model_iv1, model_iv2), dv)
NPSstimcue_groupwise <- summarySEwithin(
data = NPSstimcue_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
NPSstimcue_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
combined_se_calc_cooksd <-
rbind(combined_se_calc_cooksd, NPSstimcue_groupwise)
# plot parameters ______________________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
subject <- "sub"
ggtitle <- paste(taskname, " - NPS Cooksd removed")
title <- paste(taskname, " - NPS")
xlab <- ""
ylab <- "NPS (degree)"
ylim <- c(-1,1)
dv_keyword <- "NPS"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c("#4274AD", "#C5263A")
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), "_cooksd.png",
sep = ""
)
)
### Lineplots with only low cue ____________________________________
subsetNPS <- NPSstimcue_groupwise[NPSstimcue_groupwise$cue_ordered == "low cue",]
g <- cueR::plot_lineplot_twofactor_subset(subsetNPS, taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = "#4274AD", "#D73027"),
ggtitle = dv_title,
xlab = "Stimulus intensity", ylab = "Mean score")
g + theme(aspect.ratio=.8)
### Lineplots________________________________________________________________________
g <- cueR::plot_lineplot_twofactor_subset(NPSstimcue_groupwise, taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = "#4274AD","high cue" = "#C5263A"),
ggtitle = dv_title,
xlab = "Stimulus intensity", ylab = "Mean score")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) # Adjust point size
# print(arranged_plot)
plot_filename = file.path(analysis_dir,
paste('lineplot_task-all_rating-', dv_keyword,dv, '.svg', sep = ""))
print(g)
ggsave(plot_filename, g, width = 8, height = 4, dpi=300)
}[1] "_______ 1 nps _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00538923 (tol =
0.002, component 1)[OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa 28077. 0
2 nlminbwrap 28077. 0.0000000235
3 nloptwrap.NLOPT_LN_NELDERMEAD 28077. 0.0000000510
4 optimx.L-BFGS-B 28077. 0.000000114
5 nloptwrap.NLOPT_LN_BOBYQA 28077. 0.000000525
6 nmkbw 28077. 0.000000719
7 Nelder_Mead 28077. 0.0000136 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
[1] "_______ 2 npspos _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00654829 (tol =
0.002, component 1)[OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : Warning in optimx.check(par, optcfg$ufn, optcfg$ugr, optcfg$uhess, lower, : Parameters or bounds appear to have different scalings.
This can cause poor performance in optimization.
It is important for derivative free methods like BOBYQA, UOBYQA, NEWUOA.[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa 22819. 0
2 optimx.L-BFGS-B 22819. 0.000000129
3 nlminbwrap 22819. 0.000000298
4 nloptwrap.NLOPT_LN_BOBYQA 22819. 0.000000769
5 nloptwrap.NLOPT_LN_NELDERMEAD 22819. 0.00000153
6 nmkbw 22819. 0.00000172
7 Nelder_Mead 22819. 0.0000170 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 3 npsneg _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"Warning: Model failed to converge with 1 negative eigenvalue: -1.4e+00bobyqa : [OK]
Nelder_Mead : Warning: Model failed to converge with 1 negative eigenvalue: -1.9e+00[OK]
nlminbwrap : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00213547 (tol =
0.002, component 1)[OK]
nmkbw : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00255455 (tol =
0.002, component 1)[OK]
optimx.L-BFGS-B : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0143562 (tol =
0.002, component 1)[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : Warning: Model failed to converge with 1 negative eigenvalue: -1.7e+00[OK]
nloptwrap.NLOPT_LN_BOBYQA : Warning: Model failed to converge with 1 negative eigenvalue: -1.4e+00[OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 nloptwrap.NLOPT_LN_NELDERMEAD 14927. 0
2 bobyqa 14927. 0.0000000160
3 nlminbwrap 14927. 0.00000444
4 nmkbw 14927. 0.0000128
5 optimx.L-BFGS-B 14927. 0.0000816
6 nloptwrap.NLOPT_LN_BOBYQA 14927. 0.0128
7 Nelder_Mead 14927. 0.0129 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 4 pos_nps_vermis _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : Warning: Model failed to converge with 1 negative eigenvalue: -6.8e+00[OK]
Nelder_Mead : [OK]
nlminbwrap : Warning: Model failed to converge with 1 negative eigenvalue: -6.8e+01[OK]
nmkbw : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00207468 (tol =
0.002, component 1)[OK]
optimx.L-BFGS-B : Warning: Model failed to converge with 1 negative eigenvalue: -5.3e+01[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 Nelder_Mead -13100. 0
2 nloptwrap.NLOPT_LN_BOBYQA -13100. 0.0000000322
3 nloptwrap.NLOPT_LN_NELDERMEAD -13100. 0.0000000364
4 nmkbw -13100. 0.00000118
5 bobyqa -13099. 0.187
6 nlminbwrap -13099. 0.187
7 optimx.L-BFGS-B -13099. 0.187 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 5 pos_nps_rIns _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0666718 (tol =
0.002, component 1)[OK]
nlminbwrap : [OK]
nmkbw : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0023069 (tol =
0.002, component 1)[OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa 8729. 0
2 optimx.L-BFGS-B 8729. 0.00000000739
3 nloptwrap.NLOPT_LN_NELDERMEAD 8729. 0.0000000821
4 nloptwrap.NLOPT_LN_BOBYQA 8729. 0.000000116
5 nlminbwrap 8729. 0.000000158
6 nmkbw 8729. 0.00000164
7 Nelder_Mead 8729. 0.00119 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 6 pos_nps_rV1 _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00414318 (tol =
0.002, component 1)[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -2581. 0
2 Nelder_Mead -2581. 1.86e-10
3 nlminbwrap -2581. 2.42e-10
4 nloptwrap.NLOPT_LN_NELDERMEAD -2581. 1.28e- 8
5 nloptwrap.NLOPT_LN_BOBYQA -2581. 7.80e- 8
6 nmkbw -2581. 4.40e- 7
7 optimx.L-BFGS-B -2581. 8.56e- 6Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 7 pos_nps_rThal _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : Warning: Model failed to converge with 1 negative eigenvalue: -3.5e+02[OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -8600. 0
2 nlminbwrap -8600. 0.00000000108
3 nloptwrap.NLOPT_LN_BOBYQA -8600. 0.0000000169
4 Nelder_Mead -8600. 0.000000420
5 optimx.L-BFGS-B -8600. 0.000000441
6 nloptwrap.NLOPT_LN_NELDERMEAD -8600. 0.000000567
7 nmkbw -8600. 0.00000164 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 8 pos_nps_lIns _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : Warning in optimx.check(par, optcfg$ufn, optcfg$ugr, optcfg$uhess, lower, : Parameters or bounds appear to have different scalings.
This can cause poor performance in optimization.
It is important for derivative free methods like BOBYQA, UOBYQA, NEWUOA.[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -3622. 0
2 nlminbwrap -3622. 5.68e-10
3 Nelder_Mead -3622. 1.52e- 9
4 nloptwrap.NLOPT_LN_NELDERMEAD -3622. 4.18e- 8
5 nmkbw -3622. 3.26e- 7
6 nloptwrap.NLOPT_LN_BOBYQA -3622. 3.42e- 7
7 optimx.L-BFGS-B -3622. 1.10e- 6Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 9 pos_nps_rdpIns _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : Warning: Model failed to converge with 1 negative eigenvalue: -2.5e+01[OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00498826 (tol =
0.002, component 1)[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 nloptwrap.NLOPT_LN_NELDERMEAD -8983. 0
2 Nelder_Mead -8983. 0.00000000850
3 nlminbwrap -8983. 0.0000000213
4 nloptwrap.NLOPT_LN_BOBYQA -8983. 0.0000000347
5 nmkbw -8983. 0.000000859
6 optimx.L-BFGS-B -8983. 0.00000797
7 bobyqa -8983. 0.0406 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 10 pos_nps_rS2_Op _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -2146. 0
2 nlminbwrap -2146. 0.00000000522
3 Nelder_Mead -2146. 0.00000000859
4 nloptwrap.NLOPT_LN_NELDERMEAD -2146. 0.0000000540
5 optimx.L-BFGS-B -2146. 0.0000000741
6 nloptwrap.NLOPT_LN_BOBYQA -2146. 0.000000187
7 nmkbw -2146. 0.000000397 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 11 pos_nps_dACC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa 9557. 0
2 nlminbwrap 9557. 0.0000000136
3 nloptwrap.NLOPT_LN_NELDERMEAD 9557. 0.000000173
4 nmkbw 9557. 0.000000976
5 nloptwrap.NLOPT_LN_BOBYQA 9557. 0.00000119
6 optimx.L-BFGS-B 9557. 0.00000352
7 Nelder_Mead 9557. 0.0000124 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 12 neg_nps_rLOC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : Warning: Model failed to converge with 1 negative eigenvalue: -2.2e+01[OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -4664. 0
2 Nelder_Mead -4664. 6.88e-10
3 nloptwrap.NLOPT_LN_NELDERMEAD -4664. 1.21e- 8
4 nlminbwrap -4664. 5.24e- 8
5 nloptwrap.NLOPT_LN_BOBYQA -4664. 1.56e- 7
6 optimx.L-BFGS-B -4664. 5.99e- 7
7 nmkbw -4664. 3.62e- 3Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 13 neg_nps_lLOC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : Warning: Model failed to converge with 1 negative eigenvalue: -8.4e+02[OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -7689. 0
2 Nelder_Mead -7689. 0.00000000595
3 nloptwrap.NLOPT_LN_NELDERMEAD -7689. 0.0000000247
4 nloptwrap.NLOPT_LN_BOBYQA -7689. 0.0000000560
5 nmkbw -7689. 0.000000281
6 optimx.L-BFGS-B -7689. 0.000000419
7 nlminbwrap -7685. 1.90 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 14 neg_nps_rpLOC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"Warning: Model failed to converge with 1 negative eigenvalue: -4.6e-01bobyqa : Warning: Model failed to converge with 1 negative eigenvalue: -7.1e+01[OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : Warning: Model failed to converge with 1 negative eigenvalue: -8.2e+01[OK]
nloptwrap.NLOPT_LN_BOBYQA : Warning: Model failed to converge with 1 negative eigenvalue: -4.6e-01[OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 nlminbwrap -1573. 0
2 Nelder_Mead -1573. 4.76e-10
3 nmkbw -1573. 1.38e- 6
4 optimx.L-BFGS-B -1573. 1.48e- 6
5 bobyqa -1573. 1.50e- 2
6 nloptwrap.NLOPT_LN_NELDERMEAD -1573. 1.50e- 2
7 nloptwrap.NLOPT_LN_BOBYQA -1573. 1.50e- 2Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 15 neg_nps_pgACC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0371812 (tol =
0.002, component 1)[OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -7616. 0
2 nlminbwrap -7616. 3.08e-10
3 optimx.L-BFGS-B -7616. 4.85e- 9
4 nloptwrap.NLOPT_LN_BOBYQA -7616. 1.63e- 8
5 nloptwrap.NLOPT_LN_NELDERMEAD -7616. 4.84e- 8
6 nmkbw -7616. 6.52e- 7
7 Nelder_Mead -7616. 3.47e- 4Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 16 neg_nps_lSTS _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -2.4e+00bobyqa : Warning: Model failed to converge with 1 negative eigenvalue: -2.9e+00[OK]
Nelder_Mead : [OK]
nlminbwrap : Warning: Model failed to converge with 1 negative eigenvalue: -3.0e+00[OK]
nmkbw : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.0353021 (tol =
0.002, component 1)[OK]
optimx.L-BFGS-B : Warning: Model failed to converge with 1 negative eigenvalue: -9.4e-01[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -2.4e+00[OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 nmkbw -2550. 0
2 nloptwrap.NLOPT_LN_BOBYQA -2550. 0.000917
3 bobyqa -2550. 0.000990
4 optimx.L-BFGS-B -2550. 0.000990
5 nlminbwrap -2550. 0.000990
6 Nelder_Mead -2550. 0.000990
7 nloptwrap.NLOPT_LN_NELDERMEAD -2550. 0.000990Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
[1] "_______ 17 neg_nps_rIPL _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : [OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -8648. 0
2 optimx.L-BFGS-B -8648. 0.00000000182
3 nlminbwrap -8648. 0.00000000747
4 Nelder_Mead -8648. 0.0000000228
5 nloptwrap.NLOPT_LN_BOBYQA -8648. 0.0000000534
6 nloptwrap.NLOPT_LN_NELDERMEAD -8648. 0.0000000768
7 nmkbw -8648. 0.000000263 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
[1] "_______ 18 neg_nps_PCC _______"
[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=92"
bobyqa : [OK]
Nelder_Mead : Warning: Model failed to converge with 1 negative eigenvalue: -1.6e-01[OK]
nlminbwrap : [OK]
nmkbw : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00236044 (tol =
0.002, component 1)[OK]
optimx.L-BFGS-B : [OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]
# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa -3803. 0
2 nlminbwrap -3803. 0.00000000220
3 optimx.L-BFGS-B -3803. 0.0000000487
4 nloptwrap.NLOPT_LN_NELDERMEAD -3803. 0.0000000900
5 nloptwrap.NLOPT_LN_BOBYQA -3803. 0.000000199
6 nmkbw -3803. 0.00000160
7 Nelder_Mead -3803. 0.287 Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`
Ignoring unknown parameters: `width`
5 SIIPS (df_filter.NA)
SIIPS linear analysis
model.SIIPS <- lmer(SIIPS ~ CUE_high_gt_low*STIM_linear +
CUE_high_gt_low*STIM_quadratic + (CUE_high_gt_low*STIM_linear|sub),
data = df_filter.NA,
control=lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
summary(model.SIIPS)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: SIIPS ~ CUE_high_gt_low * STIM_linear + CUE_high_gt_low * STIM_quadratic + (CUE_high_gt_low * STIM_linear | sub)
Data: df_filter.NA
Control: lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-04))
REML criterion at convergence: 74772.9
Scaled residuals:
Min 1Q Median 3Q Max
-6.1802 -0.5584 0.0115 0.5581 5.0361
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 72328.3 268.94
CUE_high_gt_low 1312.0 36.22 0.23
STIM_linear 318.8 17.85 0.97 0.00
CUE_high_gt_low:STIM_linear 2478.5 49.78 1.00 0.15 0.99
Residual 159652.9 399.57
Number of obs: 5029, groups: sub, 92
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 578.047 28.700 90.785 20.141 < 2e-16 ***
CUE_high_gt_low 35.484 11.948 86.226 2.970 0.00386 **
STIM_linear 116.723 13.938 1478.815 8.374 < 2e-16 ***
STIM_quadratic 33.905 12.068 4852.260 2.809 0.00498 **
CUE_high_gt_low:STIM_linear -39.572 28.113 1010.404 -1.408 0.15955
CUE_high_gt_low:STIM_quadratic 3.901 24.136 4853.001 0.162 0.87159
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) CUE_h__ STIM_l STIM_q CUE_hgh_gt_lw:STIM_l
CUE_hgh_gt_ 0.072
STIM_linear 0.128 0.005
STIM_qudrtc 0.000 -0.001 -0.002
CUE_hgh_gt_lw:STIM_l 0.182 0.010 0.026 -0.004
CUE_hgh_gt_lw:STIM_q 0.000 -0.001 -0.004 0.000 -0.002 sjPlot::tab_model(model.SIIPS,
title = "Multilevel-modeling: \nlmer(SIIPS ~ CUE * STIM + (CUE + STIM | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| SIIPS | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 578.05 | 521.78 – 634.31 | <0.001 |
| CUE high gt low | 35.48 | 12.06 – 58.91 | 0.003 |
| STIM linear | 116.72 | 89.40 – 144.05 | <0.001 |
| STIM quadratic | 33.91 | 10.25 – 57.56 | 0.005 |
|
CUE high gt low × STIM linear |
-39.57 | -94.69 – 15.54 | 0.159 |
|
CUE high gt low × STIM quadratic |
3.90 | -43.42 – 51.22 | 0.872 |
| Random Effects | |||
| σ2 | 159652.92 | ||
| τ00 sub | 72328.33 | ||
| τ11 sub.CUE_high_gt_low | 1312.04 | ||
| τ11 sub.STIM_linear | 318.80 | ||
| τ11 sub.CUE_high_gt_low:STIM_linear | 2478.54 | ||
| ρ01 | 0.23 | ||
| 0.97 | |||
| 1.00 | |||
| N sub | 92 | ||
| Observations | 5029 | ||
| Marginal R2 / Conditional R2 | 0.018 / NA | ||
sink("/Users/h/Desktop/SIIPS_output.txt") # Redirects output to 'my_output.txt'
summary(model.SIIPS)equatiomatic::extract_eq(model.SIIPS)\[ \begin{aligned} \operatorname{SIIPS}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2j[i]}(\operatorname{STIM\_linear}) + \beta_{3}(\operatorname{STIM\_quadratic}) + \beta_{4j[i]}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{5}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{2j} \\ &\beta_{4j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{2j}} \\ &\mu_{\beta_{4j}} \end{aligned} \end{array} \right) , \left( \begin{array}{cccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{2j}} & \rho_{\alpha_{j}\beta_{4j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{2j}} & \rho_{\beta_{1j}\beta_{4j}} \\ \rho_{\beta_{2j}\alpha_{j}} & \rho_{\beta_{2j}\beta_{1j}} & \sigma^2_{\beta_{2j}} & \rho_{\beta_{2j}\beta_{4j}} \\ \rho_{\beta_{4j}\alpha_{j}} & \rho_{\beta_{4j}\beta_{1j}} & \rho_{\beta_{4j}\beta_{2j}} & \sigma^2_{\beta_{4j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
sink(NULL) #dv <- "SIIPS"
taskname <- "pain"
subject <- "sub"
df_filter$stim_name <- NA; df_filter$cue_name <- NA
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <- "low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
SIIPS_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2), dv)
SIIPS_groupwise <- summarySEwithin(
data = SIIPS_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2),
idvar = subject
)
SIIPS_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "SIIPS"
ylim <- c(-10, 60)
dv_keyword <- "siips"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)
SIIPSsubset <- SIIPS_groupwise[SIIPS_groupwise$cue_ordered == "low cue",]
g <- plot_lineplot_twofactor(SIIPSsubset,
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex), ggtitle = paste0("NPS N =", length(unique(SIIPS_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "SIIPS")
g <- g + theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim(465, 650)
ggsave(file.path(analysis_dir, paste0("SIIPS_maineffect", as.character(Sys.Date()), "_lowcue.svg")), plot=g, w=4, h=4, dpi=300)
k <- cueR::plot_lineplot_twofactor_subset(SIIPS_groupwise, taskname = "pain",
iv1 = "stim_ordered", iv2 = "cue_ordered",
mean = "mean_per_sub_norm_mean", error = "se",
color = c("low cue" = lowcuehex,"high cue" = highcuehex),
ggtitle = paste0("SIIPS N =", length(unique(SIIPS_subjectwise$sub))),
xlab = "Stimulus intensity", ylab = "SIIPS")Warning in geom_pointrange(aes(ymin = (.data[[mean]] - .data[[error]]), : Ignoring unknown parameters: `width`k<- k+ theme(aspect.ratio=1,
text = element_text(size = 18), # Default text size for the plot
axis.title = element_text(size = 22, ), # Axis titles
axis.text = element_text(size = 15), # Axis text (x and y)
plot.title = element_text(size = 24, hjust = 0.5) # Plot title
) +
# geom_line(size = 1) + # Adjust line thickness
geom_point(size = 2) + # Adjust point size
ylim(465, 650)
k
ggsave(file.path(analysis_dir, paste0("SIIPS_maineffect", as.character(Sys.Date()), ".svg")), plot=k, w=4, h=4, dpi=300)5-3 SIIPS and PE (PEmerge)
SIIPSPE <- lmer(SIIPS ~ CUE_high_gt_low + STIM_linear + PE_mdl2 + (CUE_high_gt_low + STIM_linear |sub),
data = PEmerge,
control=lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
summary(SIIPSPE)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: SIIPS ~ CUE_high_gt_low + STIM_linear + PE_mdl2 + (CUE_high_gt_low + STIM_linear | sub)
Data: PEmerge
Control: lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-04))
REML criterion at convergence: 73434.7
Scaled residuals:
Min 1Q Median 3Q Max
-6.1080 -0.5520 0.0123 0.5535 5.0921
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 71713.1 267.79
CUE_high_gt_low 1739.6 41.71 0.29
STIM_linear 258.9 16.09 0.96 0.00
Residual 159922.1 399.90
Number of obs: 4937, groups: sub, 88
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 583.7659 29.2568 87.5302 19.953 < 2e-16 ***
CUE_high_gt_low 59.5861 17.4640 216.3324 3.412 0.00077 ***
STIM_linear 93.0478 18.2732 2012.3596 5.092 3.87e-07 ***
PE_mdl2 0.8563 0.4149 1181.2445 2.064 0.03926 *
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation of Fixed Effects:
(Intr) CUE___ STIM_l
CUE_hgh_gt_ 0.028
STIM_linear 0.129 -0.452
PE_mdl2 -0.064 0.710 -0.639vif(SIIPSPE)CUE_high_gt_low STIM_linear PE_mdl2
2.019107 1.689254 2.715155 Session wise filter
subjects_with_inadequate_data <- data %>%
group_by(sub, CUE_high_gt_low, STIM_linear, SES_linear) %>% #SES_linear,
dplyr::summarise(count = n(), .groups = 'drop') %>%
filter(count < 4) %>%
distinct(sub) %>%
pull(sub)
df_filter_ses <- data %>%
filter(!(sub %in% subjects_with_inadequate_data))
print(sprintf("after filtering out subjects that have less than 3 trials in cell, we have N=%d -> N=%d",length(unique(data$sub)), length(unique(df_filter_ses$sub)) ))[1] "after filtering out subjects that have less than 3 trials in cell, we have N=97 -> N=44"4 Sessionwise Beh
model.behses <- lmer(outcomerating_impute ~
CUE_high_gt_low*STIM_linear*SES_linear +
CUE_high_gt_low*STIM_quadratic*SES_linear +
CUE_high_gt_low*STIM_linear*SES_quadratic +
CUE_high_gt_low*STIM_quadratic*SES_quadratic +
(CUE_high_gt_low + STIM_linear + SES_linear |
sub), data = df_filter_ses,
control=lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-4)))
# control = lmerControl(optimizer = "nmkbw"
# ))
summary(model.behses)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: outcomerating_impute ~ CUE_high_gt_low * STIM_linear * SES_linear +
CUE_high_gt_low * STIM_quadratic * SES_linear + CUE_high_gt_low *
STIM_linear * SES_quadratic + CUE_high_gt_low * STIM_quadratic * SES_quadratic + (CUE_high_gt_low + STIM_linear + SES_linear |
sub)
Data: df_filter_ses
Control: lmerControl(check.conv.singular = .makeCC(action = "ignore", tol = 1e-04))
REML criterion at convergence: 20555.8
Scaled residuals:
Min 1Q Median 3Q Max
-4.8057 -0.5810 0.0106 0.5651 4.2667
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 603.02 24.556
CUE_high_gt_low 67.63 8.224 -0.30
STIM_linear 82.33 9.074 0.58 -0.22
SES_linear 1663.11 40.781 -0.14 0.07 -0.09
Residual 292.13 17.092
Number of obs: 2372, groups: sub, 44
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 57.1311 3.8879 40.4750 14.695 < 2e-16 ***
CUE_high_gt_low 6.6941 1.4487 38.9592 4.621 4.12e-05 ***
STIM_linear 28.8428 1.6423 41.5695 17.562 < 2e-16 ***
SES_linear -12.0518 6.7229 36.9439 -1.793 0.0812 .
STIM_quadratic 1.4593 0.7521 2189.6397 1.940 0.0525 .
SES_quadratic -1.8859 0.9102 2247.4062 -2.072 0.0384 *
CUE_high_gt_low:STIM_linear -1.7178 1.7200 2189.6986 -0.999 0.3180
CUE_high_gt_low:SES_linear -1.2866 1.8565 1774.2405 -0.693 0.4884
STIM_linear:SES_linear 1.6410 2.2438 1696.3550 0.731 0.4647
CUE_high_gt_low:STIM_quadratic -3.0132 1.5043 2189.6209 -2.003 0.0453 *
SES_linear:STIM_quadratic -2.6821 1.8427 2189.6425 -1.456 0.1457
CUE_high_gt_low:SES_quadratic 0.3361 1.5593 2183.1024 0.216 0.8294
STIM_linear:SES_quadratic -3.3002 1.8979 2156.6074 -1.739 0.0822 .
STIM_quadratic:SES_quadratic 1.9730 1.6113 2189.6373 1.224 0.2209
CUE_high_gt_low:STIM_linear:SES_linear 5.2783 4.2117 2189.7226 1.253 0.2103
CUE_high_gt_low:SES_linear:STIM_quadratic 5.3577 3.6854 2189.6425 1.454 0.1462
CUE_high_gt_low:STIM_linear:SES_quadratic 2.2848 3.6868 2189.6742 0.620 0.5355
CUE_high_gt_low:STIM_quadratic:SES_quadratic -1.4627 3.2226 2189.5944 -0.454 0.6500
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation matrix not shown by default, as p = 18 > 12.
Use print(x, correlation=TRUE) or
vcov(x) if you need itsjPlot::tab_model(model.behses,
title = "Multilevel-modeling: \nlmer(BEH ~ CUE * STIM *SES + (CUE + STIM + SES | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| outcomerating_impute | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 57.13 | 49.51 – 64.76 | <0.001 |
| CUE high gt low | 6.69 | 3.85 – 9.54 | <0.001 |
| STIM linear | 28.84 | 25.62 – 32.06 | <0.001 |
| SES linear | -12.05 | -25.24 – 1.13 | 0.073 |
| STIM quadratic | 1.46 | -0.02 – 2.93 | 0.052 |
| SES quadratic | -1.89 | -3.67 – -0.10 | 0.038 |
|
CUE high gt low × STIM linear |
-1.72 | -5.09 – 1.66 | 0.318 |
|
CUE high gt low × SES linear |
-1.29 | -4.93 – 2.35 | 0.488 |
| STIM linear × SES linear | 1.64 | -2.76 – 6.04 | 0.465 |
|
CUE high gt low × STIM quadratic |
-3.01 | -5.96 – -0.06 | 0.045 |
|
SES linear × STIM quadratic |
-2.68 | -6.30 – 0.93 | 0.146 |
|
CUE high gt low × SES quadratic |
0.34 | -2.72 – 3.39 | 0.829 |
|
STIM linear × SES quadratic |
-3.30 | -7.02 – 0.42 | 0.082 |
|
STIM quadratic × SES quadratic |
1.97 | -1.19 – 5.13 | 0.221 |
|
(CUE high gt low × STIM linear) × SES linear |
5.28 | -2.98 – 13.54 | 0.210 |
|
(CUE high gt low × SES linear) × STIM quadratic |
5.36 | -1.87 – 12.58 | 0.146 |
|
(CUE high gt low × STIM linear) × SES quadratic |
2.28 | -4.94 – 9.51 | 0.535 |
|
(CUE high gt low × STIM quadratic) × SES quadratic |
-1.46 | -7.78 – 4.86 | 0.650 |
| Random Effects | |||
| σ2 | 292.13 | ||
| τ00 sub | 603.02 | ||
| τ11 sub.CUE_high_gt_low | 67.63 | ||
| τ11 sub.STIM_linear | 82.33 | ||
| τ11 sub.SES_linear | 1663.11 | ||
| ρ01 | -0.30 | ||
| 0.58 | |||
| -0.14 | |||
| ICC | 0.76 | ||
| N sub | 44 | ||
| Observations | 2372 | ||
| Marginal R2 / Conditional R2 | 0.129 / 0.788 | ||
equatiomatic::extract_eq(model.behses)\[ \begin{aligned} \operatorname{outcomerating\_impute}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2j[i]}(\operatorname{STIM\_linear}) + \beta_{3j[i]}(\operatorname{SES\_linear}) + \beta_{4}(\operatorname{STIM\_quadratic}) + \beta_{5}(\operatorname{SES\_quadratic}) + \beta_{6}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{7}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear}) + \beta_{8}(\operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{9}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) + \beta_{10}(\operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{11}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic}) + \beta_{12}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{13}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) + \beta_{14}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{15}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{16}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{17}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{2j} \\ &\beta_{3j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{2j}} \\ &\mu_{\beta_{3j}} \end{aligned} \end{array} \right) , \left( \begin{array}{cccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{2j}} & \rho_{\alpha_{j}\beta_{3j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{2j}} & \rho_{\beta_{1j}\beta_{3j}} \\ \rho_{\beta_{2j}\alpha_{j}} & \rho_{\beta_{2j}\beta_{1j}} & \sigma^2_{\beta_{2j}} & \rho_{\beta_{2j}\beta_{3j}} \\ \rho_{\beta_{3j}\alpha_{j}} & \rho_{\beta_{3j}\beta_{1j}} & \rho_{\beta_{3j}\beta_{2j}} & \sigma^2_{\beta_{3j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
# Check why convergence is negative ____________________________________________
abeh <- lme4::allFit(model.behses)bobyqa : [OK]
Nelder_Mead : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 1.13484 (tol =
0.002, component 1)[OK]
nlminbwrap : [OK]
nmkbw : [OK]
optimx.L-BFGS-B : Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00784696 (tol =
0.002, component 1)[OK]
nloptwrap.NLOPT_LN_NELDERMEAD : [OK]
nloptwrap.NLOPT_LN_BOBYQA : [OK]opt.summary <- glance(abeh) |> select(optimizer, AIC, NLL_rel) |> arrange(NLL_rel)
print(opt.summary)# A tibble: 7 × 3
optimizer AIC NLL_rel
<chr> <dbl> <dbl>
1 bobyqa 20614. 0
2 nlminbwrap 20614. 0.0000000733
3 nloptwrap.NLOPT_LN_NELDERMEAD 20614. 0.000000199
4 nmkbw 20614. 0.000000468
5 nloptwrap.NLOPT_LN_BOBYQA 20614. 0.00000132
6 optimx.L-BFGS-B 20614. 0.0000431
7 Nelder_Mead 20615. 0.687 5 Sessionwise NPS
model.NPSses <- lmer(NPS ~
CUE_high_gt_low*STIM_linear*SES_linear +
CUE_high_gt_low*STIM_quadratic*SES_linear +
CUE_high_gt_low*STIM_linear*SES_quadratic +
CUE_high_gt_low*STIM_quadratic*SES_quadratic +
(CUE_high_gt_low + STIM_linear +SES_linear |
sub), data = df_filter_ses,
control = lmerControl(
optimizer = "optimx",
optCtrl = list(
method = "nmkb",
maxit = 1e9,
maxfun = 1e9,
maxeval = 1e7,
xtol_abs = 1e-9,
ftol_abs = 1e-9
)
)
)Warning in ctrl[namc] <- control: number of items to replace is not a multiple of replacement lengthWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -8.6e-03summary(model.NPSses)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: NPS ~ CUE_high_gt_low * STIM_linear * SES_linear + CUE_high_gt_low *
STIM_quadratic * SES_linear + CUE_high_gt_low * STIM_linear * SES_quadratic + CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(CUE_high_gt_low + STIM_linear + SES_linear | sub)
Data: df_filter_ses
Control: lmerControl(optimizer = "optimx", optCtrl = list(method = "nmkb",
maxit = 1e+09, maxfun = 1e+09, maxeval = 1e+07, xtol_abs = 1e-09, ftol_abs = 1e-09))
REML criterion at convergence: 12378.5
Scaled residuals:
Min 1Q Median 3Q Max
-5.9321 -0.5543 -0.0081 0.5686 5.3204
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 2.300694 1.51680
CUE_high_gt_low 0.139566 0.37359 -0.57
STIM_linear 0.004609 0.06789 0.34 0.58
SES_linear 4.796639 2.19012 -0.16 -0.68 -0.93
Residual 10.020358 3.16549
Number of obs: 2376, groups: sub, 44
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 4.73676 0.24820 43.52579 19.085 <2e-16 ***
CUE_high_gt_low -0.03354 0.14242 43.21367 -0.236 0.8149
STIM_linear 1.33863 0.15947 1305.94556 8.394 <2e-16 ***
SES_linear 0.11153 0.40678 32.94520 0.274 0.7857
STIM_quadratic 0.38680 0.13919 2237.43650 2.779 0.0055 **
SES_quadratic 0.27111 0.15777 1639.77499 1.718 0.0859 .
CUE_high_gt_low:STIM_linear -0.02950 0.31824 2237.43650 -0.093 0.9262
CUE_high_gt_low:SES_linear -0.08578 0.32062 1489.12394 -0.268 0.7891
STIM_linear:SES_linear 0.32546 0.38989 2239.84736 0.835 0.4039
CUE_high_gt_low:STIM_quadratic -0.05087 0.27839 2237.43650 -0.183 0.8550
SES_linear:STIM_quadratic 0.05290 0.34101 2237.43650 0.155 0.8767
CUE_high_gt_low:SES_quadratic 0.44446 0.27919 2111.56958 1.592 0.1115
STIM_linear:SES_quadratic -0.12938 0.34093 2241.90566 -0.380 0.7044
STIM_quadratic:SES_quadratic 0.43692 0.29821 2237.43650 1.465 0.1430
CUE_high_gt_low:STIM_linear:SES_linear 0.21128 0.77965 2237.43650 0.271 0.7864
CUE_high_gt_low:SES_linear:STIM_quadratic 0.13414 0.68202 2237.43650 0.197 0.8441
CUE_high_gt_low:STIM_linear:SES_quadratic 0.34457 0.68181 2237.43650 0.505 0.6133
CUE_high_gt_low:STIM_quadratic:SES_quadratic 0.22317 0.59643 2237.43650 0.374 0.7083
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation matrix not shown by default, as p = 18 > 12.
Use print(x, correlation=TRUE) or
vcov(x) if you need itoptimizer (optimx) convergence code: 0 (OK)
unable to evaluate scaled gradient
Model failed to converge: degenerate Hessian with 1 negative eigenvalues
number of items to replace is not a multiple of replacement lengthsjPlot::tab_model(model.NPSses,
title = "Multilevel-modeling: \nlmer(NPS ~ CUE * STIM * SES+ (CUE + STIM + SES| sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| NPS | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 4.74 | 4.25 – 5.22 | <0.001 |
| CUE high gt low | -0.03 | -0.31 – 0.25 | 0.814 |
| STIM linear | 1.34 | 1.03 – 1.65 | <0.001 |
| SES linear | 0.11 | -0.69 – 0.91 | 0.784 |
| STIM quadratic | 0.39 | 0.11 – 0.66 | 0.005 |
| SES quadratic | 0.27 | -0.04 – 0.58 | 0.086 |
|
CUE high gt low × STIM linear |
-0.03 | -0.65 – 0.59 | 0.926 |
|
CUE high gt low × SES linear |
-0.09 | -0.71 – 0.54 | 0.789 |
| STIM linear × SES linear | 0.33 | -0.44 – 1.09 | 0.404 |
|
CUE high gt low × STIM quadratic |
-0.05 | -0.60 – 0.50 | 0.855 |
|
SES linear × STIM quadratic |
0.05 | -0.62 – 0.72 | 0.877 |
|
CUE high gt low × SES quadratic |
0.44 | -0.10 – 0.99 | 0.112 |
|
STIM linear × SES quadratic |
-0.13 | -0.80 – 0.54 | 0.704 |
|
STIM quadratic × SES quadratic |
0.44 | -0.15 – 1.02 | 0.143 |
|
(CUE high gt low × STIM linear) × SES linear |
0.21 | -1.32 – 1.74 | 0.786 |
|
(CUE high gt low × SES linear) × STIM quadratic |
0.13 | -1.20 – 1.47 | 0.844 |
|
(CUE high gt low × STIM linear) × SES quadratic |
0.34 | -0.99 – 1.68 | 0.613 |
|
(CUE high gt low × STIM quadratic) × SES quadratic |
0.22 | -0.95 – 1.39 | 0.708 |
| Random Effects | |||
| σ2 | 10.02 | ||
| τ00 sub | 2.30 | ||
| τ11 sub.CUE_high_gt_low | 0.14 | ||
| τ11 sub.STIM_linear | 0.00 | ||
| τ11 sub.SES_linear | 4.80 | ||
| ρ01 | -0.57 | ||
| 0.34 | |||
| -0.16 | |||
| ICC | 0.24 | ||
| N sub | 44 | ||
| Observations | 2376 | ||
| Marginal R2 / Conditional R2 | 0.028 / 0.259 | ||
equatiomatic::extract_eq(model.NPSses)\[ \begin{aligned} \operatorname{NPS}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2j[i]}(\operatorname{STIM\_linear}) + \beta_{3j[i]}(\operatorname{SES\_linear}) + \beta_{4}(\operatorname{STIM\_quadratic}) + \beta_{5}(\operatorname{SES\_quadratic}) + \beta_{6}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{7}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear}) + \beta_{8}(\operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{9}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) + \beta_{10}(\operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{11}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic}) + \beta_{12}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{13}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) + \beta_{14}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{15}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{16}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{17}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{2j} \\ &\beta_{3j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{2j}} \\ &\mu_{\beta_{3j}} \end{aligned} \end{array} \right) , \left( \begin{array}{cccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{2j}} & \rho_{\alpha_{j}\beta_{3j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{2j}} & \rho_{\beta_{1j}\beta_{3j}} \\ \rho_{\beta_{2j}\alpha_{j}} & \rho_{\beta_{2j}\beta_{1j}} & \sigma^2_{\beta_{2j}} & \rho_{\beta_{2j}\beta_{3j}} \\ \rho_{\beta_{3j}\alpha_{j}} & \rho_{\beta_{3j}\beta_{1j}} & \rho_{\beta_{3j}\beta_{2j}} & \sigma^2_{\beta_{3j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
optimx_options <- c("L-BFGS-B", "nlminb", "nlm", "bobyqa", "nmkb", "hjkb")
for(i in 1:length(optimx_options)){
print(paste0("Testing optimx: ", optimx_options[i]))
model_flex <- lmer(
NPS ~
CUE_high_gt_low * STIM_linear * SES_linear +
CUE_high_gt_low * STIM_quadratic * SES_linear +
CUE_high_gt_low * STIM_linear * SES_quadratic +
CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(CUE_high_gt_low + SES_linear + SES_quadratic + STIM_linear + STIM_quadratic |
sub),
data = df_filter_ses,
control = lmerControl(
optimizer = "optimx",
optCtrl = list(
method = optimx_options[i],
maxit = 1e9,
maxfun = 1e9,
maxeval = 1e7,
xtol_abs = 1e-9,
ftol_abs = 1e-9
)
)
)
if(is.null(model_flex@optinfo$conv$lme4$messages)){
print(paste0("One of the optimx options, ", optimx_options[i],", worked!"))
print(model_flex)
break
}
}[1] "Testing optimx: L-BFGS-B"Warning in optim(par = par, fn = ufn, gr = ugr, lower = lower, upper = upper, : unknown names in control: maxfun, maxeval, xtol_abs,
ftol_absboundary (singular) fit: see help('isSingular')[1] "Testing optimx: nlminb"Warning in nlminb(start = par, objective = ufn, gradient = ugr, lower = lower, : unrecognized control elements named 'maxfun', 'maxeval',
'xtol_abs', 'ftol_abs' ignoredboundary (singular) fit: see help('isSingular')[1] "Testing optimx: nlm"Warning in nlminb(start = par, objective = ufn, gradient = ugr, lower = lower, : unrecognized control elements named 'maxfun', 'maxeval',
'xtol_abs', 'ftol_abs' ignoredboundary (singular) fit: see help('isSingular')[1] "Testing optimx: bobyqa"Warning in (function (npt = min(n + 2L, 2L * n), rhobeg = NA, rhoend = NA, : unused control arguments ignoredboundary (singular) fit: see help('isSingular')[1] "Testing optimx: nmkb"Warning in ctrl[namc] <- control: number of items to replace is not a multiple of replacement lengthboundary (singular) fit: see help('isSingular')[1] "Testing optimx: hjkb"Warning in cntrl[nmsCo] <- control: number of items to replace is not a multiple of replacement lengthstep nofc fmin xpar
1 44 12618.31 1 ...
2 429 12368 0.5 ...
3 543 12367.21 0.5 ...
4 657 12360.07 0.5 ...
5 848 12358.46 0.5 ...
6 1196 12355.68 0.46875 ...
7 1549 12355.43 0.453125 ...
8 1749 12355.36 0.4609375 ...
9 1950 12355.34 0.4570312 ...
10 2151 12355.33 0.4570312 ...
11 3104 12355.33 0.4580078 ...
12 3456 12355.33 0.4580078 ...
13 3730 12355.33 0.4577637 ...
14 4004 12355.33 0.4577637 ...
15 4122 12355.33 0.4577637 ...
16 4240 12355.33 0.4577637 ...
17 4436 12355.33 0.4577484 ...
18 4632 12355.33 0.4577484 ...
19 4906 12355.33 0.4577484 ...boundary (singular) fit: see help('isSingular')Warning: Model failed to converge with 2 negative eigenvalues: -6.1e-02 -9.7e+006 Sessionwise SIIPS
random slopes of cue effects were colinear with random intercepts with a correlation of 1, therefore estimating the idnvidual difference baseline and dindividual difference of cue effects were impossible
with(df_filter_ses, table(CUE_high_gt_low, STIM_linear, SES_linear)), , SES_linear = -0.5
STIM_linear
CUE_high_gt_low -0.5 0 0.5
-0.5 128 128 128
0.5 128 128 128
, , SES_linear = 0
STIM_linear
CUE_high_gt_low -0.5 0 0.5
-0.5 132 132 132
0.5 132 132 132
, , SES_linear = 0.5
STIM_linear
CUE_high_gt_low -0.5 0 0.5
-0.5 136 136 136
0.5 136 136 136library(glmmTMB)
model.SIIPSses <- lmer(SIIPS ~
CUE_high_gt_low*STIM_linear*SES_linear +
CUE_high_gt_low*STIM_quadratic*SES_linear +
CUE_high_gt_low*STIM_linear*SES_quadratic +
CUE_high_gt_low*STIM_quadratic*SES_quadratic +
(1+ CUE_high_gt_low + STIM_quadratic + SES_linear|sub),
data = df_filter_ses,
control = lmerControl(optimizer = "nmkbw"
))Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00464962 (tol =
0.002, component 1)summary(model.SIIPSses)Linear mixed model fit by REML. t-tests use Satterthwaite's method ['lmerModLmerTest']
Formula: SIIPS ~ CUE_high_gt_low * STIM_linear * SES_linear + CUE_high_gt_low *
STIM_quadratic * SES_linear + CUE_high_gt_low * STIM_linear * SES_quadratic + CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(1 + CUE_high_gt_low + STIM_quadratic + SES_linear | sub)
Data: df_filter_ses
Control: lmerControl(optimizer = "nmkbw")
REML criterion at convergence: 34296.1
Scaled residuals:
Min 1Q Median 3Q Max
-4.0754 -0.5766 0.0244 0.5720 5.2361
Random effects:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 60450 245.87
CUE_high_gt_low 1888 43.45 0.13
STIM_quadratic 7729 87.92 0.39 0.08
SES_linear 33645 183.43 -0.24 0.23 -0.94
Residual 107187 327.39
Number of obs: 2376, groups: sub, 44
Fixed effects:
Estimate Std. Error df t value Pr(>|t|)
(Intercept) 545.354 38.223 42.294 14.268 < 2e-16 ***
CUE_high_gt_low 14.105 15.111 35.145 0.933 0.3570
STIM_linear 121.136 16.457 2188.708 7.361 2.58e-13 ***
SES_linear 8.258 36.270 25.450 0.228 0.8217
STIM_quadratic 20.103 19.854 37.825 1.013 0.3177
SES_quadratic 8.918 16.001 1364.870 0.557 0.5774
CUE_high_gt_low:STIM_linear -19.787 32.915 2188.708 -0.601 0.5478
CUE_high_gt_low:SES_linear -18.557 33.374 1322.426 -0.556 0.5783
STIM_linear:SES_linear 30.159 40.318 2188.708 0.748 0.4545
CUE_high_gt_low:STIM_quadratic -5.768 28.793 2188.708 -0.200 0.8412
SES_linear:STIM_quadratic -35.822 36.313 1439.624 -0.986 0.3241
CUE_high_gt_low:SES_quadratic 30.691 28.975 2025.616 1.059 0.2896
STIM_linear:SES_quadratic 18.933 35.258 2188.708 0.537 0.5913
STIM_quadratic:SES_quadratic 60.722 31.247 2071.576 1.943 0.0521 .
CUE_high_gt_low:STIM_linear:SES_linear 43.819 80.636 2188.708 0.543 0.5869
CUE_high_gt_low:SES_linear:STIM_quadratic 66.566 70.538 2188.708 0.944 0.3454
CUE_high_gt_low:STIM_linear:SES_quadratic -130.658 70.517 2188.708 -1.853 0.0640 .
CUE_high_gt_low:STIM_quadratic:SES_quadratic 33.497 61.686 2188.708 0.543 0.5872
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Correlation matrix not shown by default, as p = 18 > 12.
Use print(x, correlation=TRUE) or
vcov(x) if you need itoptimizer (nmkbw) convergence code: 0 (OK)
Model failed to converge with max|grad| = 0.00464962 (tol = 0.002, component 1)sjPlot::tab_model(model.SIIPSses,
title = "Multilevel-modeling: \nlmer(SIIPS ~ CUE * STIM * SES+ (CUE + STIM *SES | sub), data = pain)",
CSS = list(css.table = '+font-size: 12;'))| SIIPS | |||
|---|---|---|---|
| Predictors | Estimates | CI | p |
| (Intercept) | 545.35 | 470.40 – 620.31 | <0.001 |
| CUE high gt low | 14.10 | -15.53 – 43.74 | 0.351 |
| STIM linear | 121.14 | 88.86 – 153.41 | <0.001 |
| SES linear | 8.26 | -62.87 – 79.38 | 0.820 |
| STIM quadratic | 20.10 | -18.83 – 59.04 | 0.311 |
| SES quadratic | 8.92 | -22.46 – 40.30 | 0.577 |
|
CUE high gt low × STIM linear |
-19.79 | -84.33 – 44.76 | 0.548 |
|
CUE high gt low × SES linear |
-18.56 | -84.00 – 46.89 | 0.578 |
| STIM linear × SES linear | 30.16 | -48.90 – 109.22 | 0.455 |
|
CUE high gt low × STIM quadratic |
-5.77 | -62.23 – 50.69 | 0.841 |
|
SES linear × STIM quadratic |
-35.82 | -107.03 – 35.39 | 0.324 |
|
CUE high gt low × SES quadratic |
30.69 | -26.13 – 87.51 | 0.290 |
|
STIM linear × SES quadratic |
18.93 | -50.21 – 88.07 | 0.591 |
|
STIM quadratic × SES quadratic |
60.72 | -0.55 – 122.00 | 0.052 |
|
(CUE high gt low × STIM linear) × SES linear |
43.82 | -114.31 – 201.94 | 0.587 |
|
(CUE high gt low × SES linear) × STIM quadratic |
66.57 | -71.76 – 204.89 | 0.345 |
|
(CUE high gt low × STIM linear) × SES quadratic |
-130.66 | -268.94 – 7.62 | 0.064 |
|
(CUE high gt low × STIM quadratic) × SES quadratic |
33.50 | -87.47 – 154.46 | 0.587 |
| Random Effects | |||
| σ2 | 107187.41 | ||
| τ00 sub | 60449.67 | ||
| τ11 sub.CUE_high_gt_low | 1888.25 | ||
| τ11 sub.STIM_quadratic | 7729.14 | ||
| τ11 sub.SES_linear | 33644.96 | ||
| ρ01 | 0.13 | ||
| 0.39 | |||
| -0.24 | |||
| ICC | 0.39 | ||
| N sub | 44 | ||
| Observations | 2376 | ||
| Marginal R2 / Conditional R2 | 0.018 / 0.399 | ||
equatiomatic::extract_eq(model.SIIPSses)\[ \begin{aligned} \operatorname{SIIPS}_{i} &\sim N \left(\mu, \sigma^2 \right) \\ \mu &=\alpha_{j[i]} + \beta_{1j[i]}(\operatorname{CUE\_high\_gt\_low}) + \beta_{2}(\operatorname{STIM\_linear}) + \beta_{3j[i]}(\operatorname{SES\_linear}) + \beta_{4j[i]}(\operatorname{STIM\_quadratic}) + \beta_{5}(\operatorname{SES\_quadratic}) + \beta_{6}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_linear}) + \beta_{7}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear}) + \beta_{8}(\operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{9}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{STIM\_quadratic}) + \beta_{10}(\operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{11}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic}) + \beta_{12}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{13}(\operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) + \beta_{14}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_linear}) + \beta_{15}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_linear} \times \operatorname{STIM\_quadratic}) + \beta_{16}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_linear}) + \beta_{17}(\operatorname{CUE\_high\_gt\_low} \times \operatorname{SES\_quadratic} \times \operatorname{STIM\_quadratic}) \\ \left( \begin{array}{c} \begin{aligned} &\alpha_{j} \\ &\beta_{1j} \\ &\beta_{3j} \\ &\beta_{4j} \end{aligned} \end{array} \right) &\sim N \left( \left( \begin{array}{c} \begin{aligned} &\mu_{\alpha_{j}} \\ &\mu_{\beta_{1j}} \\ &\mu_{\beta_{3j}} \\ &\mu_{\beta_{4j}} \end{aligned} \end{array} \right) , \left( \begin{array}{cccc} \sigma^2_{\alpha_{j}} & \rho_{\alpha_{j}\beta_{1j}} & \rho_{\alpha_{j}\beta_{3j}} & \rho_{\alpha_{j}\beta_{4j}} \\ \rho_{\beta_{1j}\alpha_{j}} & \sigma^2_{\beta_{1j}} & \rho_{\beta_{1j}\beta_{3j}} & \rho_{\beta_{1j}\beta_{4j}} \\ \rho_{\beta_{3j}\alpha_{j}} & \rho_{\beta_{3j}\beta_{1j}} & \sigma^2_{\beta_{3j}} & \rho_{\beta_{3j}\beta_{4j}} \\ \rho_{\beta_{4j}\alpha_{j}} & \rho_{\beta_{4j}\beta_{1j}} & \rho_{\beta_{4j}\beta_{3j}} & \sigma^2_{\beta_{4j}} \end{array} \right) \right) \text{, for sub j = 1,} \dots \text{,J} \end{aligned} \]
Session is inversly colinear with stimulus intensity, which makes the random slopes impossible to estimate
library(ggplot2); theme_set(theme_bw())
library(ggalt)
(ggplot(df_filter_ses, aes(STIM_linear,SIIPS,SES_linear,colour=CUE_high_gt_low))
+geom_point()
+geom_encircle(aes(group=sub))
+theme(legend.position="none")
+ scale_y_log10()
)Warning in transformation$transform(x): NaNs producedWarning in scale_y_log10(): log-10 transformation introduced infinite values.Warning in transformation$transform(x): NaNs producedWarning in scale_y_log10(): log-10 transformation introduced infinite values.Warning: Removed 159 rows containing missing values or values outside the scale range (`geom_point()`).Warning: Removed 159 rows containing missing values or values outside the scale range (`geom_encircle()`).
linear estimator
optimx_options <- c("L-BFGS-B", "nlminb", "nlm", "bobyqa", "nmkb", "hjkb")
for(i in 1:length(optimx_options)){
print(paste0("Testing optimx: ", optimx_options[i]))
model_flex <- lmer(
SIIPS ~
CUE_high_gt_low * STIM_linear * SES_linear +
CUE_high_gt_low * STIM_quadratic * SES_linear +
CUE_high_gt_low * STIM_linear * SES_quadratic +
CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(1+ CUE_high_gt_low + STIM_quadratic + SES_linear |
sub),
data = df_filter_ses,
control = lmerControl(
optimizer = "optimx",
optCtrl = list(
method = optimx_options[i],
maxit = 1e9,
maxfun = 1e9,
maxeval = 1e3,
xtol_abs = 1e-3,
ftol_abs = 1e-3
)
)
)
if(is.null(model_flex@optinfo$conv$lme4$messages)){
print(paste0("One of the optimx options, ", optimx_options[i],", worked!"))
print(model_flex)
break
}
}[1] "Testing optimx: L-BFGS-B"Warning in optim(par = par, fn = ufn, gr = ugr, lower = lower, upper = upper, : unknown names in control: maxfun, maxeval, xtol_abs,
ftol_absWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.00679432 (tol =
0.002, component 1)[1] "Testing optimx: nlminb"Warning in nlminb(start = par, objective = ufn, gradient = ugr, lower = lower, : unrecognized control elements named 'maxfun', 'maxeval',
'xtol_abs', 'ftol_abs' ignored[1] "One of the optimx options, nlminb, worked!"
Linear mixed model fit by REML ['lmerModLmerTest']
Formula: SIIPS ~ CUE_high_gt_low * STIM_linear * SES_linear + CUE_high_gt_low *
STIM_quadratic * SES_linear + CUE_high_gt_low * STIM_linear * SES_quadratic + CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(1 + CUE_high_gt_low + STIM_quadratic + SES_linear | sub)
Data: df_filter_ses
REML criterion at convergence: 34296.09
Random effects:
Groups Name Std.Dev. Corr
sub (Intercept) 245.91
CUE_high_gt_low 43.44 0.13
STIM_quadratic 87.92 0.39 0.08
SES_linear 183.42 -0.24 0.23 -0.94
Residual 327.39
Number of obs: 2376, groups: sub, 44
Fixed Effects:
(Intercept) CUE_high_gt_low STIM_linear
545.355 14.103 121.136
SES_linear STIM_quadratic SES_quadratic
8.267 20.104 8.921
CUE_high_gt_low:STIM_linear CUE_high_gt_low:SES_linear STIM_linear:SES_linear
-19.787 -18.558 30.159
CUE_high_gt_low:STIM_quadratic SES_linear:STIM_quadratic CUE_high_gt_low:SES_quadratic
-5.768 -35.824 30.689
STIM_linear:SES_quadratic STIM_quadratic:SES_quadratic CUE_high_gt_low:STIM_linear:SES_linear
18.933 60.720 43.819
CUE_high_gt_low:SES_linear:STIM_quadratic CUE_high_gt_low:STIM_linear:SES_quadratic CUE_high_gt_low:STIM_quadratic:SES_quadratic
66.566 -130.658 33.497
optimizer (optimx) convergence code: 0 (OK) ; 1 optimizer warnings; 0 lme4 warnings # non linear options
algoptions <- c("NLOPT_LN_PRAXIS", "NLOPT_GN_CRS2_LM",
"NLOPT_LN_COBYLA", "NLOPT_LN_NEWUOA",
"NLOPT_LN_NEWUOA_BOUND", "NLOPT_LN_NELDERMEAD",
"NLOPT_LN_SBPLX", "NLOPT_LN_BOBYQA")
for(i in 1:length(algoptions)){
print(paste0("Testing nloptwrap: ", algoptions[i]))
model_flex <- lmer( SIIPS ~
CUE_high_gt_low * STIM_linear * SES_linear +
CUE_high_gt_low * STIM_quadratic * SES_linear +
CUE_high_gt_low * STIM_linear * SES_quadratic +
CUE_high_gt_low * STIM_quadratic * SES_quadratic +
( 1 +CUE_high_gt_low + STIM_quadratic + SES_linear |
sub),
data = df_filter_ses, REML=FALSE,
control = lmerControl(optimizer = "nloptwrap",
optCtrl = list(algorithm = algoptions[i],
maxit = 1e9,
maxfun = 1e9,
maxeval = 1e9,
xtol_abs = 1e-2,
ftol_abs = 1e-2)))
if(is.null(model_flex@optinfo$conv$lme4$messages)){
print(paste0("One of the nloptwrap options, ", algoptions[i],", worked!"))
print(model_flex)
break
}
}[1] "Testing nloptwrap: NLOPT_LN_PRAXIS"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -1.0e+01[1] "Testing nloptwrap: NLOPT_GN_CRS2_LM"Warning in optwrap(optimizer, devfun, getStart(start, rho$pp), lower = rho$lower, : convergence code -2 from nloptwrap: NLOPT_INVALID_ARGS:
Invalid arguments (e.g. lower bounds are bigger than upper bounds, an unknown algorithm was specified, etcetera).Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 3 negative
eigenvaluesWarning: Model failed to converge with 3 negative eigenvalues: -4.8e-01 -3.4e+01 -4.7e+01[1] "Testing nloptwrap: NLOPT_LN_COBYLA"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.322768 (tol =
0.002, component 1)[1] "Testing nloptwrap: NLOPT_LN_NEWUOA"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 1.37552 (tol =
0.002, component 1)[1] "Testing nloptwrap: NLOPT_LN_NEWUOA_BOUND"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -2.2e+01[1] "Testing nloptwrap: NLOPT_LN_NELDERMEAD"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge with max|grad| = 0.283849 (tol =
0.002, component 1)[1] "Testing nloptwrap: NLOPT_LN_SBPLX"boundary (singular) fit: see help('isSingular')[1] "Testing nloptwrap: NLOPT_LN_BOBYQA"Warning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : unable to evaluate scaled gradientWarning in checkConv(attr(opt, "derivs"), opt$par, ctrl = control$checkConv, : Model failed to converge: degenerate Hessian with 1 negative
eigenvaluesWarning: Model failed to converge with 1 negative eigenvalue: -4.9e+01 model.glmm <- glmmTMB::glmmTMB( SIIPS ~
CUE_high_gt_low * STIM_linear * SES_linear +
CUE_high_gt_low * STIM_quadratic * SES_linear +
CUE_high_gt_low * STIM_linear * SES_quadratic +
CUE_high_gt_low * STIM_quadratic * SES_quadratic +
( CUE_high_gt_low + SES_linear + STIM_linear | sub),
data = df_filter_ses)Warning in finalizeTMB(TMBStruc, obj, fit, h, data.tmb.old): Model convergence problem; non-positive-definite Hessian matrix. See
vignette('troubleshooting')summary(model.glmm) Family: gaussian ( identity )
Formula: SIIPS ~ CUE_high_gt_low * STIM_linear * SES_linear + CUE_high_gt_low *
STIM_quadratic * SES_linear + CUE_high_gt_low * STIM_linear * SES_quadratic + CUE_high_gt_low * STIM_quadratic * SES_quadratic +
(CUE_high_gt_low + SES_linear + STIM_linear | sub)
Data: df_filter_ses
AIC BIC logLik deviance df.resid
NA NA NA NA 2347
Random effects:
Conditional model:
Groups Name Variance Std.Dev. Corr
sub (Intercept) 5.496e+04 2.344e+02
CUE_high_gt_low 4.959e-10 2.227e-05 1.00
SES_linear 1.737e-20 1.318e-10 -0.89 -0.87
STIM_linear 2.643e-15 5.141e-08 -0.79 -0.77 0.98
Residual 1.114e+05 3.338e+02
Number of obs: 2376, groups: sub, 44
Dispersion estimate for gaussian family (sigma^2): 1.11e+05
Conditional model:
Estimate Std. Error z value Pr(>|z|)
(Intercept) 551.6641 36.1360 15.266 < 2e-16 ***
CUE_high_gt_low 15.5396 13.7004 1.134 0.2567
STIM_linear 121.1456 16.7795 7.220 5.2e-13 ***
SES_linear -8.1260 18.9826 -0.428 0.6686
STIM_quadratic 23.8181 14.6783 1.623 0.1047
SES_quadratic 0.9281 15.5296 0.060 0.9523
CUE_high_gt_low:STIM_linear -19.9400 33.5590 -0.594 0.5524
CUE_high_gt_low:SES_linear -17.6618 33.5642 -0.526 0.5987
STIM_linear:SES_linear 29.2349 41.1076 0.711 0.4770
CUE_high_gt_low:STIM_quadratic -5.5709 29.3566 -0.190 0.8495
SES_linear:STIM_quadratic -47.7892 35.9598 -1.329 0.1839
CUE_high_gt_low:SES_quadratic 29.0409 29.3520 0.989 0.3225
STIM_linear:SES_quadratic 18.6771 35.9488 0.520 0.6034
STIM_quadratic:SES_quadratic 57.0768 31.4470 1.815 0.0695 .
CUE_high_gt_low:STIM_linear:SES_linear 40.6604 82.2151 0.495 0.6209
CUE_high_gt_low:SES_linear:STIM_quadratic 66.4087 71.9196 0.923 0.3558
CUE_high_gt_low:STIM_linear:SES_quadratic -130.8349 71.8975 -1.820 0.0688 .
CUE_high_gt_low:STIM_quadratic:SES_quadratic 34.8856 62.8941 0.555 0.5791
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1dv <- "SIIPS"
taskname <- "pain"
subject <- "sub"
# Plot 1: reordering variables for plots _______________________________________
df_filter$cue_name[df_filter$cue == "high_cue"] <- "high cue"
df_filter$cue_name[df_filter$cue == "low_cue"] <- "low cue"
df_filter$stim_name[df_filter$stimulusintensity == "high_stim"] <- "High"
df_filter$stim_name[df_filter$stimulusintensity == "med_stim"] <- "Med"
df_filter$stim_name[df_filter$stimulusintensity == "low_stim"] <- "Low"
df_filter$stim_ordered <- factor(df_filter$stim_name, levels = c("Low", "Med", "High"))
df_filter$cue_ordered <- factor(df_filter$cue_name, levels = c("low cue", "high cue"))
model_iv1 <- "stim_ordered"
model_iv2 <- "cue_ordered"
# Plot 2: summary statistics __________________________________________________
SIIPS_subjectwise <- meanSummary(df_filter,
c(subject, model_iv1, model_iv2, "ses"), dv)
SIIPS_groupwise <- summarySEwithin(
data = SIIPS_subjectwise,
measurevar = "mean_per_sub",
withinvars = c(model_iv1, model_iv2, "ses"),
idvar = subject
)Automatically converting the following non-factors to factors: sesSIIPS_groupwise$task <- taskname
# https://stackoverflow.com/questions/29402528/append-data-frames-together-in-a-for-loop/29419402
# Plot 3: plot parameters _____________________________________________________
sub_mean <- "mean_per_sub"
group_mean <- "mean_per_sub_norm_mean"
se <- "se"
ggtitle <- paste(taskname, " - beh Cooksd removed")
title <- paste(taskname, " - beh")
xlab <- ""
ylab <- "SIIPS"
ylim <- c(-10, 60)
dv_keyword <- "siips"
if (any(startsWith(dv_keyword, c("expect", "Expect")))) {
color <- c("#1B9E77", "#D95F02")
} else {
color <- c(lowcuehex, highcuehex)
} # if keyword starts with
plot_savefname <- file.path(
analysis_dir,
paste(
"raincloud_task-", taskname, "_rating-", dv_keyword, "_", as.character(Sys.Date()), ".png",
sep = ""
)
)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